U.S. patent number 5,754,714 [Application Number 08/526,384] was granted by the patent office on 1998-05-19 for semiconductor optical waveguide device, optical control type optical switch, and wavelength conversion device.
This patent grant is currently assigned to Kabushiki Kaisha Toshiba. Invention is credited to Yuzo Hirayama, Nobuo Suzuki.
United States Patent |
5,754,714 |
Suzuki , et al. |
May 19, 1998 |
Semiconductor optical waveguide device, optical control type
optical switch, and wavelength conversion device
Abstract
A semiconductor optical waveguide device comprises a
stripe-shaped semiconductor optical waveguide, part of the
semiconductor optical waveguide being an active layer producing
gain by electric current injection, and part of the semiconductor
optical waveguide being an intra-band resonant absorption layer in
which an intra-band absorption resonant wavelength is arranged
within the gain band of the active layer, and means for injecting
electric current into the active layer.
Inventors: |
Suzuki; Nobuo (Tokyo,
JP), Hirayama; Yuzo (Yokohama, JP) |
Assignee: |
Kabushiki Kaisha Toshiba
(Kawasaki, JP)
|
Family
ID: |
27291037 |
Appl.
No.: |
08/526,384 |
Filed: |
September 11, 1995 |
Foreign Application Priority Data
|
|
|
|
|
Sep 17, 1994 [JP] |
|
|
6-248451 |
Mar 1, 1995 [JP] |
|
|
7-042014 |
Apr 26, 1995 [JP] |
|
|
7-102198 |
|
Current U.S.
Class: |
385/5; 385/131;
385/16 |
Current CPC
Class: |
B82Y
20/00 (20130101); G02F 2/004 (20130101); H01S
5/0601 (20130101); G02F 1/01708 (20130101); G02F
1/01716 (20130101); G02F 1/3133 (20130101); G02F
1/3517 (20130101); H01S 5/0602 (20130101); H01S
5/0608 (20130101); H01S 5/0609 (20130101); H01S
5/0611 (20130101); H01S 5/50 (20130101); H01S
5/5054 (20130101) |
Current International
Class: |
G02F
2/00 (20060101); H01S 5/06 (20060101); H01S
5/00 (20060101); G02F 1/01 (20060101); G02F
1/017 (20060101); G02F 1/29 (20060101); G02F
1/313 (20060101); G02F 1/35 (20060101); H01S
5/50 (20060101); G02F 001/295 () |
Field of
Search: |
;385/5,4,8,9,14,15,16,3,131 ;372/102 |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Fifth Optoelectronics Conference (Dec. '94) Technical Digest, Paper
13E2-2; "Light-Controlled Optical Modulation Using Three Energy
Levels in Undoped Quantum Well Structure", Susumu Noda, et al. pp.
92-93, Jul. 1994. .
Electron Intersubband Transistions to 0.8 eV (1.55 .mu.m) in
InGaAs/AlAs Single Quantum Wells, by J. H. Smet, L. H. Peng, Y.
Hirayama, and C. G. Fonstad; Appl. Phys. Lett. vol. 64, No. 8, Feb.
21, 1994. .
Feasibility of 1.55 .mu.m Intersubband Photonic Devices Using
InGaAs/AlAs Pseudomorphic Quantum Well Structures, Yuzo Hirayama,
Jurgen H. Smet, Lung-Han Peng, Clifton G. Fonstad and Erich P.
Ippen, Jpn. J. Appl., Phys., vol. 33 (1994), pp. 890-895 Jan.
1994..
|
Primary Examiner: Palmer; Phan T. H.
Attorney, Agent or Firm: Oblon, Spivak, McClelland, Maier
& Neustadt, P.C.
Claims
What is claimed is:
1. A semiconductor optical waveguide device comprising:
a stripe-shaped semiconductor optical waveguide, part of said
semiconductor optical waveguide being an active layer producing
gain by electric current injection, and part of said semiconductor
optical waveguide being an intra-band resonant absorption layer in
which an intra-band absorption resonant wavelength is arranged
within the gain band of said active layer; and
means for injecting electric current into said active layer.
2. A semiconductor optical waveguide device according to claim 1,
wherein said intra-band resonant absorption layer and said active
layer closely laminated and constitute an integral optical
waveguide.
3. A semiconductor optical waveguide device according to claim 2,
wherein said intra-band resonant absorption layer is made of a
material in which a bandgap is greater than double of that of said
active layer.
4. A semiconductor optical waveguide device according to claim 3,
wherein said active layer is composed of one or more III-V group
compound semiconductors mainly including arsenic element and/or
phosphorus element, and said intra-band resonant absorption layer
has a multi-layer structure and is composed of III-V group compound
semiconductors mainly including nitrogen element.
5. A semiconductor optical waveguide device according to claim 2,
further comprising means for controlling resonant wavelength of
said intra-band resonant absorption layer.
6. A semiconductor optical waveguide device according to claim 5,
wherein said means for controlling resonant wavelength of said
intra-band resonant absorption layer is divided into a plurality of
sections.
7. A semiconductor optical waveguide device according to claim 5,
wherein impurities are doped into at least one part of said
intra-band resonant absorption layer.
8. A semiconductor optical waveguide device according to claim 1,
wherein the intra-band absorption resonant wavelength of said
intra-band resonant absorption layer is an inter-valence band
absorption resonant wavelength.
9. A semiconductor optical waveguide device according to claim 1,
wherein said intra-band resonant absorption layer has a quantum
well structure, and its intra-band absorption resonant wavelength
is an inter-subband transition resonant wavelength of the quantum
well.
10. A semiconductor optical waveguide device according to claim 1,
wherein said intra-band resonant absorption layer is identical to
said active layer.
11. A semiconductor optical waveguide device according to claim 1,
wherein control light and signal light are introduced into said
semiconductor optical waveguide and the control light modifies
intensity and/or phase of the signal light, and density of
electrons and holes injected into said active layer are regulated
so as to establish an equilibrium between loss and gain of said
semiconductor optical waveguide at and near the wavelength of the
control light and the signal light.
12. A semiconductor optical waveguide device according to claim 1,
wherein lights having a plurality of wavelength is introduced into
said semiconductor optical waveguide, and another light having
wavelength different from any one of that of lights is generated by
four wave mixing in said semiconductor optical waveguide.
13. An optical control type optical switch in which at least one of
destination, intensity, wavelength, and phase of a signal light
output is controlled by a control light, comprising:
a stripe-shaped semiconductor optical waveguide, part of said
semiconductor optical waveguide being an active layer;
means for inputting and outputting the signal light and the control
light; and
means for regulating the density of electrons and holes of the
active layer so as to establish an equilibrium between loss and
gain of said semiconductor optical waveguide at and near the
wavelength of the signal light and the control light,
wherein part of said semiconductor optical waveguide is made of a
material having an intra-band resonant absorption at the wavelength
of the control light.
14. An optical control type optical switch in which at least one of
destination, intensity, wavelength, and phase of a signal light
output is controlled by a control light, comprising:
a stripe-shaped semiconductor optical waveguide constituting a part
of optical interferometer, wherein:
part of said semiconductor optical waveguide is an active layer,
said semiconductor optical waveguide has means for regulating the
density of electrons and holes of the active layer so as to
establish an equilibrium between loss and gain of said
semiconductor optical waveguide at and near the wavelength of the
signal light and the control light, and part of said semiconductor
optical waveguide is made of a material having an intra-band
resonant absorption at the wavelength of the control light.
15. A wavelength conversion device comprising:
a traveling wave type semiconductor laser amplifier having a
semiconductor optical waveguide formed by sandwiching an active
layer between a pair of clad layers, wherein light having an
angular frequency different from the lights introduced into said
semiconductor optical waveguide is generated by four wave mixing in
said semiconductor optical waveguide,
wherein said semiconductor optical waveguide has a semiconductor
layer having an intra-band absorption resonant wavelength arranged
within the gain band of said traveling wave type semiconductor
laser amplifier.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to an optical waveguide device and, more
particularly, it relates to an optical waveguide device, an optical
control type optical switch, and a wavelength conversion device
capable of being externally controlled for intra-band transition
within its optical waveguide layer in order to operate for a
specifically assigned function.
2. Description of the Related Art
The technological development of optoelectronics in recent years
particularly in terms of semiconductor laser, low loss optical
fiber, optical fiber amplifiers and high speed integrated circuits
has made it possible to transmit data at an enhanced rate of 10
gigabits per seconds over a very long distance. In the so-called
multimedia age which we expect to see in the near future, the end
users of a data transmission network are believed to be able to
utilize a vast amount of data including highly defined visual
images on a real time basis and, for such a system to be realized,
construction of huge infrastructures that can support the operation
of high speed data transmission and processing is
indispensable.
However, despite the technological development of high speed
integrated circuits, electronic apparatuses designed to process
data at a rate greater than tens of several gigabits per second are
still very costly because of a number of problems including the
delay of data transmission over wires, a high energy consumption
rate and a high manufacturing/assembling cost. In an attempt to
bypass these problems, a new technology of optical routing that can
be used to process a vast amount of data that cannot be
electronically dealt with is attracting attention.
To put an optical routing system in place, semiconductor optical
waveguide devices (optical control type ultrahigh speed optical
routing switches (routers)) have to be developed so that switching
operations may be carried out in only several picoseconds and such
operations may be repeated for a number of times without
problem.
Most advanced optical control type ultrahigh speed optical switches
known to date may be optical switches utilizing the nonlinearity of
optical fiber and typical examples of such switches include
nonlinear optical loop mirrors and Kerr shutters. However, a
nonlinear optical switch realized by using optical fiber is
normally large and very sensitive to acoustic vibrations and hence
lacks reliable stability, not to mention a high manufacturing costs
Additionally, when such a switch is used as an optical logic
device, it shows a problem of too long delay time for each stage of
operation.
From a practical point of view, nonlinear optical switches
comprising a semiconductor optical waveguide are promising.
However, a semiconductor nonlinear optical switch is accompanied by
a problem of being poorly nonlinear, requiring too much energy for
switching operations in a non-resonant wavelength range. Although
it is highly nonlinear in a resonant wavelength range, it cannot be
operated repetitively because of a long lifetime of carriers of
electric charges and shows an enhanced degree of absorption in that
wavelength range. Therefore, no optical control type optical
switches are known to data that operate at high speed and with
improved efficiency.
The nonlinearity of active and transparent optical waveguides are
drawing attention as it may provide a solution to the above
identified problem. (See, inter alia, C. T. Hultgren, et al., Appl.
Phys. Lett., vol.59, pp.635-637, 1991, C. T. Hultgren, et al. Appl.
Phys. Lett., vol.61, pp.2767-2768, 1992.) In this regard, there is
proposed the use of a traveling wave type semiconductor laser
amplifier under a condition that the level of the bias current and
the optical wavelength are selected so as to balance the gain and
the loss.
FIGS. 1A through 1C of the accompanying drawings are cited from the
above documents to illustrate the change with time of the phase of
a pulse of transmitted probe light after passing through an excited
optical pulse (with a pulse width of 440 fs). Note that the change
with time of the phase of probe light is proportional to the change
in the internal refractive index. In FIG. 1A, the semiconductor
laser amplifier is biased to the gain side and the gain is apt to
become saturated as strong exciting light is applied because
carriers are consumed at an enhanced rate to amplify the incident
light. If, on the other hand, the semiconductor laser amplifier is
biased to the loss side as shown in FIG. 1C, the loss is apt to
become saturated by carriers excited by a strongly excited optical
pulse. In either case, the device is restored from a saturated
condition with a period of time that corresponds to the carrier
lifetime. Thus, a number of factors that require several
nanoseconds for recovering from saturation can simultaneously
affect the change in the refractive index to make the device unable
to operate stably and repetitively at high speed.
With a transparent condition of FIG. 1B, to the contrary, neither
gain saturation nor loss saturation takes place and hence no slow
factors appear on the change in the refractive index so that only
quick changes in the refractive index can be utilized.
Additionally, since any loss is compensated by the stimulated
emission gain given rise to by current injection, the insertion
loss can be suppressed to a low level to make multi-stage
connection highly feasible.
A quick change in the refractive index can be divided in to a
negative change component that shows a large initial value and a
positive change component that takes place immediately after the
negative change. The initial negative change in the refractive
index is assumedly caused by excitation of carriers that takes
place as a result of two-photon absorption (TPA) and/or free
carrier absorption. A carrier excited to a high energy state loses
its energy within 1 picosecond through the collision with phonons
and another carriers to relax to a low energy state. It is believed
that a positive change occurs in the refractive index as the
average temperature of carriers of electric charges rises (carrier
heating) during the energy transition. A heated carrier further
loses its energy as it collides with phonons to return to its
original state within several picoseconds. Thus, if only such rapid
changes in the refractive index can be utilized, a high speed
repetitive operation may be realized at a rate of several hundred
gigabits per second.
Two-photon absorption is believed to be mainly attributable to
nonlinearity in a non-resonant wavelength range. While no
significant nonlinearity is usually achievable in a non-resonant
wavelength range, nonlinearity can be realized to a satisfactory
level in this instance because of the fact that incident light is
in a resonant wavelength range. In a tentative calculation using
some of the values shown in an available research document (K. L.
Hall et al., Appl. Phys. Lett., vol.62, pp.1320-1322, 1933), a peak
power approximately expressed by formula 5.2W/L will give rise to a
shift of .pi. to probe light for an exciting pulse having a pulse
width greater than 1 picosecond, where L is the device length
expressed in terms of millimeters. In other words, a switching
operation can be realized at a peak power of about 500 mW by using
a device having a length of 10 mm. Although such a peak power is
actually attainable with a currently available semiconductor pulse
laser, the peak power level will have to be reduced for practical
applications.
Thus, it may appear at the first glance that a high speed optical
switching operation can be realized by means of an active
transparent optical waveguide. However, in reality, since electrons
are excited from a valence band to a conduction band as a result of
two-photon absorption, the actual speed of operation is restricted
by the carrier accumulation. More specifically, while electrons and
holes excited to a high energy level as a result of two-photon
absorption may lose part of their energy and become relaxed as they
fall close to the bottom of the conduction band or the top of the
valence band within a short period of time of several picoseconds,
they are still excessively excited in that state. While only few
carriers may be excited by a single pulse, there may be given rise
to a large number of excited carriers as the pulse is repeated at
high speed until a saturated state is produced for absorption so
that the response may become fluctuated with a time constant
corresponding to the carrier lifetime (several nanoseconds)
depending on the pattern of excitation.
On the other hand, in an optical control type optical switch using
an active transparent optical waveguide, values close to the gain
peak wavelength may advantageously be selected for the wavelengths
of both the signal light and the control light in order to suppress
noises, although such values that are close to each other make it
difficult to separate the signal light from the control light.
Additionally a semiconductor optical control type optical switch is
accompanied by a vital drawback of being incapable of allowing a
sufficiently large extinction ratio for optical output regardless
if it comprises an active transparent optical waveguide. Now, these
problems will be discussed below by referring to a nonlinear
directional coupler (NLDC) type optical switch comprising an active
transparent optical waveguide as described in an available research
document (S. G. Lee et al., Appl. Phys. Lett., vol.64, pp.454-456,
1994).
FIG. 2 of the accompanying drawings shows a schematic cross
sectional view of a conventional NLDC type optical switch formed on
an n-type GaAs substrate 701. An active layer 701 formed on the
substrate is in fact an GaAs/AlGaAs multiple quantum well layer
which is sandwiched by an n-type AlGaAs clad layer 703 and a p-type
AlGaAs clad layer 704. A p-type GaAs contact layer 705 is formed on
the p-type AlGaAs clad layer 704. A pair of stripe-shaped mesa
regions 706a and 706b are formed on the p-type AlGaAs clad layer
704 and the p-type GaAs contact layer 705 to define respective
channels for the optical waveguide. The mesa regions have a width
of 3 .mu.m and a height of 0.9 .mu.m and are separated from each
other by a distance of 2 .mu.m. The device has an overall length of
1.3 mm. An upper electrode 707 is arranged on the upper surface of
the device including those of the mesa regions, while a lower
electrode 708 is formed under the substrate so that a substantially
transparent condition can be established for the optical input by
injecting carriers into the active layer 702.
The optical input is pulsed light having a pulse width of 200 fs
and the device is so designed as to be switched to select an output
channel depending on the energy level of the optical input
regardless of signal light or control light. The time required for
recovery is less than 1 picosecond. FIG. 3 of the accompanying
drawings shows the input energy dependency of the output ratio of
each of the channels. In FIG. 3, the broken lines are for the TE
mode and it may be appreciated that the output ratio varies between
1:3 and 1.7:1, whereas the solid lines are for the TM mode, where
the output ratio varies between 1:3 and 1.4:1. Reportedly, the
outputs of the two channels cross each other at 6 pJ for the TE
mode.
No perfect switching operation can be realized between 0:1 and 1:0
in the above instance. One of the reasons for it is that the device
length does not necessarily agree with the complete coupling length
of the directional coupler at the time of weak excitation
multiplied by a natural number. Even if such discrepancy does not
exist and an output ratio of 0:1 is realizable at the time of weak
excitation, an output ratio of 1:0 can never be achieved for an
NLDC because the equivalent refractive index at each observable
point depends on the intensity of light at that point and the
channels are coupled so that the ratio of the intensities of light
of the channels and therefore the extent of coupling at each point
change as a function of the intensity of input light to severely
damage the uniformity and the symmetry of the directional coupler
and make it substantially impossible to put the device in perfect
condition. This also holds true for an asymmetric Mach-Zehnder
interferometer type optical switch, where a large the extinction
ratio cannot be obtained at the time of strong excitation because
the waveguide performance varies at each branch point between the
time of weak excitation and that of strong excitation.
The above cited research document also describes a case where pump
light and probe light are orthogonally polarized. Control light and
signal light may be separated by such a technique. Polarization
beam splitter, however, is difficult to be integrated on a
semiconductor substrate, so that the separation of control light
and signal light is difficult on a practical basis.
The above problems may be summarized as follows.
A conventional optical control type high speed semiconductor
optical switch is not capable of carrying out switching operations
at high speed with a lower power level. Particularly, if the
optical switch is one that utilizes the nonlinearity in the active
transparent semiconductor optical waveguide, the high speed
repetitive operation capability of the device is restricted by the
lifetime of carriers accumulated by two-photon absorption (TPA).
Additionally, a conventional optical control type optical switch
cannot satisfactorily switch the destinations of optical signal
output and separate control light and signal light if quick and
highly efficient optical switching operation is sought for.
Thus, there are good reasons for expecting a high speed/high
efficiency optical control type optical switch that is not
restricted by the carrier lifetime.
Meanwhile, it may be appropriate to utilize the technology of
optical frequency-division multiplexing for transmitting and
processing a vast amount of data by optical fiber if the large
bandwidth capabilities of optical fiber is to be fully exploited.
Therefore, semiconductor optical devices that can directly change
the optical wavelength (wavelength conversion devices) without
carrying out a process of photoelectric conversion are expected to
become practically feasible in order to realize large scale and
efficient optical frequency division multiplexed networks. A
wavelength conversion device may be used for an optical
demultiplexer designed to select a signal with a specific timing
through wavelength conversion in an optical time division
multiplexed transmission system that utilizes ultrahigh speed
pulses.
A wavelength conversion device to be used for any of these
applications has to meet the requirements of a wide wavelength
conversion band, an ability of continuous wavelength conversion
within the band, response to quickly modulated signals and a high
conversion efficiency. Efforts have been and are currently being
paid to develop wavelength conversion devices that utilize the four
wave mixing in traveling wave type semiconductor laser amplifiers
for wavelength conversion in order to meet the above
requirements.
FIG. 4 of the accompanying drawing schematically shows a wavelength
conversion system comprising a conventional wavelength conversion
device that utilizes the four wave mixing in a traveling wave type
semiconductor laser amplifier for wavelength conversion. Note that
optical spectra are shown for respective stages of operation.
Referring to FIG. 4, the wavelength conversion device comprises a
semiconductor optical waveguide having a double hetero structure
and realized by sandwiching an InGaAs active layer 802 with a
p-type InP clad layer 803 and an n-type substrate 801 that also
operates as an n-type clad layer, a pair of electrodes 804, 805 for
injecting an electric current into the active layer 802 and
anti-reflection films 806 for preventing any reflection of light at
the opposite ends of the semiconductor optical waveguide.
With a wavelength conversion device having a configuration as
described above, a stimulated emission gain is given rise to as a
result of population inversion of carriers injected into the active
layer 802 at high concentration to consequently amplify any light
traveling through the optical waveguide with a wavelength found
within the gain band. Additionally, since anti-reflection films 806
are arranged at the opposite ends of the optical waveguide, any
laser oscillation is suppressed to allow optical amplification to
be realized with a large gain if a strong electric current is being
injected.
When exciting light W1 having an angular frequency of .omega.1 and
signal light W2 having an angular frequency of .omega.2 are
combined and applied to the wavelength conversion device, a change
equal to angular frequency .OMEGA.=.omega.1-.omega.2 takes place in
the optical field strength as a result of b eats produced by the
exciting light and the signal light.
Then, light with an angular frequency of .omega. will be modulated
for both intensity and phase by an angular frequency of .OMEGA. to
give rise to components of angular frequency .omega..+-.n.OMEGA.
because of the nonlinear responsiveness of gain and refractive
index relative to the internal optical field strength of the
InGaAsP active layer 802.
If it is assumed here that the power P1(0) of the exciting light is
by far greater than the power P2(0) of the signal light at the
input side end z=0, then conjugate light W3 having an angular
frequency of .omega.3 (=.omega.1+.OMEGA.) appears with the exciting
light W1 having an angular frequency of .omega.1 and the signal
light W2 with the angular frequency of .omega.2 at the output side
end z=1, and the conjugate light W3 having an angular frequency of
.omega.3 is selectively picked up by means of a narrow band optical
filter 807. Since the output power level of conjugate light is low,
it is normally amplified by an optical amplifier 808 before
use.
When the signal light is modulated for intensity or frequency, the
conjugate light is also produced in a modulated state in terms of
intensity or frequency. In other words, the light that is the
signal carrier wave is converted for wavelength, although the
spectrum of the conjugate light is inverted from that of the
original signal light.
This phenomenon can be explained by four wave mixing of
.omega.3=.omega.1+.omega.1-.omega.2, where .omega.1 may be greater
or smaller than .omega.2. The wavelength of the signal can be
converted to any value found within the band by tuning .omega.1.
According to J. Zhou, et al., IEEE Photonics Technol. Lett., vol.6,
No.1, pp.50-52, 1994, the wavelength conversion efficiency .eta. is
expressed on a dB basis by the equation below. ##EQU1## where G[dB]
is the gain of the amplifier and I.sub.p [dB] is the power of the
exciting light which is equal to 10log.sub.10 P1(0), while C.sub.m
and .tau..sub.m (m=1, 2, 3) respectively represent the complex
coupling coefficients and the time constants of three major causes
for producing four wave mixing.
The conversion efficiency .eta. is normally a large number because
it is proportional to the cube of amplification gain G and the
square of the power I.sub.p of exciting light.
The three major causes of four wave mixing are change of carrier
density (m=1), carrier heating (n=2) and spectral hole burning
(m=3).
Four wave mixing caused by change of carrier density is seen when
the number of carriers in areas having a strong optical field is
reduced by stimulated emission to reduce the gain and change the
refractive index. The time constant for it is controlled by the
effective carrier lifetime, taking stimulated emission into
consideration.
Carrier heating appears when the carrier temperature is raised as a
result of intra-band absorption of light and/or stimulated emission
to change the gain and the refractive index. The time constant for
it is controlled by the time required for carriers to become
relaxed to show the energy distribution pattern of lattice
temperature as a result of inelastic collisions and other
phenomena.
Spectral hole burning is observed when the energy distribution
pattern of carriers is shifted from Fermi distribution also as a
result of intra-band absorption of light and/or stimulated emission
to change the gain and the refractive index. The time constant for
it is controlled by the time required for carriers to become
relaxed and return to show the Fermi distribution as a result of
intercarrier collisions and other phenomena.
FIG. 5 is a graph showing the .OMEGA. dependency of the wavelength
conversion efficiency .eta. of a tensile strained InGaAs/InGaAsP
multi-quantum well traveling wave type semiconductor laser
amplifier. In the graph, the small squares are for
.omega.2>.omega.3, whereas small circles are for
.omega.2<.omega.3. It shows that wavelength conversion can be
realized over a wide range of 3.4THz (a wavelength difference of 27
nm).
By fitting the graph to the equation (1) above, C.sub.1
=0.24e.sup.-il.30 and .tau..sub.1 =200 ps, C.sub.2
=0.0027e.sup.-il.30 and .tau..sub.2 =650 fs and C.sub.3
=0.00048e.sup.-il.53 and .tau..sub.3 =50 fs are obtained.
The cutoff frequencies (wavelength differences) that correspond to
the time constants .tau..sub.1, .tau..sub.2 and .tau..sub.3 are
800MHz (0.0064 nm), 240GHz (1.9 nm) and 3.4THz (27 nm)
respectively.
The dotted line in FIG. 5 shows the slope of 20 dB/dec. If only
changes in the carrier density is involved, the wavelength
conversion efficiency .eta. falls along this line. Therefore, the
upward deviation of the conversion efficiency lines from 1 nm on is
caused by carrier heating, whereas the deviation from 20 nm on is
attributable to spectral hole burning. While carrier heating plays
a major role in the wavelength shift between 1 and 10 nm, the
conversion efficiency remains between -50 dB and -65 dB because the
absolute value of C.sub.2 is not sufficiently large.
Differently stated, while the technique of wavelength conversion by
means of a conventional wavelength conversion device utilizing the
four wave mixing effect of a known traveling wave type
semiconductor laser amplifier is effective for a conversion over a
large bandwidth of more than 10 nm of wavelength difference, thanks
to the nonlinearity of carrier heating, it is poorly effective for
a conversion over 1 nm of wavelength difference because the
nonlinearity of carrier heating is not sufficiently remarkable.
Therefore, in a wavelength conversion with a wavelength difference
greater than 1 nm, the power of conjugate light is found low
respect to the power levels of exciting light, signal light and
noises of the semiconductor laser amplifier. More specifically, the
power level of conjugate light is lower than the output power level
of signal light by about 20 dB and that of exciting light by 40 dB.
The difference between the noise level of amplified spontaneous
emission (AES) and the power level of conjugate light is somewhere
around 20 dB at most.
Thus, the extinction ratio of the narrow band optical filter 807
for picking up only conjugate light (.omega..sub.3) from the
optical output (.omega..sub.1, .omega..sub.2, .omega..sub.3) has to
be rigorously defined, although the problem of a poor signal to
noise ratio (S/N ratio) is nonetheless aggravated.
The above problems may be summarily described as follows.
Since a conventional wavelength conversion device utilizing the
four wave mixing effect of a known traveling wave type
semiconductor laser amplifier does not remarkably show nonlinearity
by carrier heating, the wavelength conversion efficiency for
greater than 1 nm of wavelength difference is rather small.
Therefore, the extinction ratio of the narrow band optical filter
of the device has to be rigorously defined, although the problem of
a poor signal to noise ratio (S/N ratio) is nonetheless
aggravated.
Therefore, there are good reasons for expecting a high efficiency
wavelength conversion device.
Meanwhile, it may be appropriate to utilize the technology of
optical frequency-division multiplexing (optical FDM) and that of
optical time-division multiplexing (optical TDM) for transmitting
and processing a vast amount of data by optical fiber if the large
bandwidth capabilities of optical fiber is to be fully exploited.
Therefore, the development of optical devices having novel
functional features is thought to be indispensable to realize large
and effective optical FDM networks and optical TDM networks.
For instance, in an optical FDM/TDM network, wavelength conversion
nodes as illustrated in FIG. 9 of the accompanying drawings are
expected to take a vital role in the signal switch and the
reutilization of wavelength channels. Referring to FIG. 9, when
signal light (wavelength .lambda..sub.q and strong exciting light
.lambda..sub.p are made to enter a wavelength conversion device
901, signal conjugate light (wavelength .lambda..sub.c) is also
produced by four wave mixing. A wavelength conversion output having
a sufficient strength can be obtained by picking up only the
.lambda..sub.c component by means of a narrow band wavelength
filter 902 and amplifying it by means of an optical fiber amplifier
903.
However, any known tunable wavelength filters can hardly switch
.lambda..sub.c at high speed. Mechanical tunable wavelength filters
and tunable wavelength filters that utilize acousto-optical effects
are too slow for switching operations and cannot feasibly be used
for the above applications. While tunable wavelength filters
realized by utilizing distributed feedback (DFB) type semiconductor
lasers and distributed Bragg reflector (DBR) type semiconductor
lasers are capable of high speed switching within, they are not
suitable as filters for short optical pulse signals because they
are of a resonant type and apt to spread the pulse width by
multiple reflections. Thus, the only feasible way of switching
.lambda..sub.q at high speed by means of any known techniques is to
feed the output of a wavelength conversion device 901 to a
plurality of narrow band optical filters and select one of the
outputs thereof and this method is an inefficient one in any sense
of the word.
It may be needless to say that the technological development of
optical FDM/TDM networks, to say nothing of that of tunable
wavelength filters, is highly dependent on novel optical devices
developed on new theories and provided with new functional
features.
Thus, there are good reasons for expecting a high efficiency
semiconductor optical waveguide device (such as a tunable
wavelength filter) that cannot be realized by conventional
techniques in order to realize optical TDM/FDM networks for the
coming multimedia age.
SUMMARY OF THE INVENTION
Therefore, an object of the present invention is to provide a high
speed/high efficiency optical control type optical switch that is
not restricted by the carrier lifetime.
Another object of the present invention is to provide an optical
control type optical switch that can easily separate control light
and signal light and can substantially perfectly select the
destination of the signal light output by switching.
Still another object of the present invention is to provide a
wavelength conversion device that operates with a conversion
efficiency higher than that of any comparable conventional
devices.
A further object of the present invention is to provide a
semiconductor optical waveguide device that can also be applied to
multifunctional light sources and light receiving devices.
As described in detail hereinafter, the essence of the first and
second aspects of the present invention lies in that, in an optical
control type optical switch that utilizes the nonlinearity of an
active transparent optical waveguide, the layered structure of the
optical waveguide is at least partly made of a material whose
resonant wavelength of intra-band absorption is substantially equal
to the wavelength of incident light so that the nonlinearity due to
intra-band absorption is made more remarkable than the nonlinearity
due to two-photon absorption.
According to the first aspect of the present invention, there is
provided an optical control type optical switch in which at least
one of destination, intensity, wavelength, and phase of a signal
light output is controlled by a control light, comprising a
stripe-shaped semiconductor optical waveguide, part of the
semiconductor optical waveguide being an active layer, means for
inputting and outputting the signal light and the control light,
and means for regulating the density of electrons and holes of the
active layer so as to establish an equilibrium between loss and
gain of the semiconductor optical waveguide at and near the
wavelength of the signal light and the control light, wherein part
of the semiconductor optical waveguide is made of a material having
an intra-band resonant wavelength substantially equal to the
wavelength of the control light.
For the purpose of the present invention, intra-band absorption
specifically refers to inter-valence band absorption from a heavy
or light hole band to a spin-orbit separation band and
inter-subband absorption in quantum well.
According to the second aspect of the present invention, there is
provided an optical control type optical switch comprising signal
light branching means for receiving signal light and branching it
to first and second intermediate optical path, a control light
input waveguide for receiving control light, a first optical
coupler for coupling the control light input waveguide and the
first intermediate optical path respectively to first and second
optical waveguides, a second optical coupler for coupling the first
and second optical waveguides respectively to a third intermediate
optical path and a control light output optical path, a reference
optical path for transmitting the light branched to the second
intermediate optical path and an output optical coupler for
coupling the third intermediate optical path and the reference
optical path respectively to first and second signal light output
optical paths, wherein the stretch between the first optical
coupler and the second optical coupler constitutes a first
Mach-Zehnder interferometer for mainly transmitting the signal
light to the third intermediate optical path regardless of the
presence or absence of control light and the phase of the signal
light transmitted to the first and second signal light output
optical path is shifted by the nonlinear optical effect of the
first and second optical waveguides, whereas the phase of the
signal light traveling through the third intermediate optical path
is shifted according to the presence or absence of control light so
as to select the destination of the principal signal light output
of the output optical coupler by switching.
Some of the preferable modes of realization of the present
invention include the following.
(1) The signal light branching means may form a Y branch or be a
1:1 optical coupler. In order for the first Mach-Zehnder
interferometer to transmit the signal light mainly to the third
intermediate optical path regardless of the presence or absence of
control light, the first and second optical couplers have to be 1:1
optical couplers and the first and second optical waveguides have
to be symmetrical relative to each other.
(2) A second Mach-Zehnder interferometer similar to the first
Mach-Zehnder interferometer has to be formed by the stretch between
the second intermediate optical path and the reference optical
path. More specifically, it is preferable that the second
Mach-Zehnder interferometer is constituted by a third optical
coupler for coupling the second control light input waveguide and
the second intermediate optical path respectively to the third and
fourth optical waveguides and a fourth optical coupler for coupling
the third and fourth optical waveguides respectively to the
reference optical path and the second control light output optical
path and inserted between the second intermediate optical path of
the output of the signal light branching means and the input of the
reference optical path of the output optical coupler in order to
provide symmetry with the optical path of the first Mach-Zehnder
interferometer.
(3) The first and second optical waveguides are constituted by
active semiconductor waveguides provided with electron current
injection means and biased so as to establish an equilibrium
between the gain and the loss relative to low-power light having
the wavelength of the control light. In an optical control type
optical switch comprising a second Mach-Zehnder interferometer as
described above, it is preferable that the third and fourth optical
waveguides are configured same as the first and second optical
waveguides respectively.
(4) The entire device is monolithically formed on a semiconductor
substrate.
(5) Means are provided to regulate the phase of the light
introduced into the second optical coupler from the first optical
waveguide and that of the light introduced into the second optical
coupler from the second optical waveguide.
(6) Means are provided to regulated the phase of the light
introduced into the output optical coupler from the reference
optical path and that of the light introduced into the output
optical coupler from the third intermediate optical path.
Carriers (electrons and holes) are present at high density in the
inside of the semiconductor optical waveguide of an optical control
type semiconductor optical switch according to the first aspect of
the invention as a result of electric current injection. Since the
optical waveguide is biased to establish an equilibrium between the
rate of carrier generation caused by interband absorption and that
of carrier loss attributable to stimulated emission for the
wavelength of exciting light. Therefore, the number of carriers in
the optical waveguide does not change significantly if exciting
light is applied there anew. On the other hand, since the
wavelength of exciting light and the resonant wavelength of
interband absorption is substantially equal to each other, part of
the carriers absorb the energy of exciting light to become excited
to a higher energy level and the carriers that are originally there
are excited within the band so that the number of carriers does not
change as a whole. As the energy distribution pattern of carriers
changes, the refractive index and the transmission coefficient
change greatly and abruptly, although excited hot carriers restore
the original equilibrium within a short period of time because of
the intra-band relaxation. In other words, although the refractive
index and the transmission coefficient can change greatly as soon
as an exciting pulse is applied to the optical waveguide, they are
restored to the respective original levels within several
picoseconds after the removal of the exciting pulse.
Strictly speaking, the carrier density is also changed slightly by
incidental two-photon absorption. However, since the intra-band
absorption is enhanced by resonance, the optical waveguide can be
used with exciting light having a power level lower than that of
its counterpart used for a conventional active transparent
waveguide to suppress the influence of two-photon absorption so
that the influence of a lasting change in the response due to
accumulation of carriers can be effectively suppressed in any quick
repetitive operation.
In an optical control type semiconductor switch according to the
second aspect of the present invention, the signal light branching
means branch the signal light to a first component fed to the first
intermediate optical path and designed to interfere with control
light and a second component fed to the second intermediate optical
path to make reference light to a ratio of 1:1. Then, the signal
light is branched by the first optical coupler to the first and
second optical waveguides to a ratio of 1:1. Furthermore, if the
control light from the first control light input waveguide is
entered in synchronism with the signal light, the control light is
also branched by the first optical coupler to the first and second
optical waveguides to a ratio of 1:1. If such control light is
present, the phase of the signal light traveling through the first
optical waveguide is shifted by .phi. as a result of nonlinear
optical effect. If the third order nonlinearity is involved, the
phase shift is proportional to the power of control light. Then,
the phase of the signal light traveling through the second optical
waveguide is also shifted by .phi..
As described above, the relationship between the phases of the two
light introduced into the second optical coupler is constant
regardless of presence or absence of control light. Since the
second optical coupler is a 1:1 coupler like the first optical
coupler, the signal light is sent to the third intermediate optical
path due to the reciprocity theorem regardless of presence or
absence of control light. On the other hand, the control light is
branched to the control light output optical paths. As a result,
the signal light and the control light are separated from each
other. Note that, since the signal light is by far weaker than the
control light, the phase shift of itself is negligible and that,
since the first and second optical couplers are passive couplers
with a small nonlinearity, the branching ratio is not substantially
affected by the presence of control light.
The phase of the signal light introduced into the output optical
coupler from the second optical coupler by way of the third
intermediate optical path is shifted by .phi. if there is control
light coming from the first control light input waveguide. On the
other hand, the phase of the reference signal light branched to the
second intermediate optical path is constant regardless of presence
or absence of control light coming from the first control light
input waveguide and the light is introduced into the output optical
coupler from the reference optical path.
Thus, the output ratio of the signal light sent from the output
optical coupler into the first output optical path to the one sent
into the second output optical path is altered by the phase of the
signal light coming from the third intermediate optical path and
the one coming from the reference optical path. If, now, it is so
designed that the output is sent to the first output optical path
in the absence of control light and the phase shift .phi. by
control light is equal to .pi. or .pi. multiplied by an odd number,
then the destination of the output is completely switched to the
second output optical path by control light. In this way, the
output ratio of signal light can be substantially perfectly
switched from 0:1 to 1:0 without the risk of mingling control light
into signal light.
However, it should be noted that the signal light of the third
intermediate optical path and that of the reference optical path
may show discrepancy in terms of intensity and pulse width to
baffle perfect switching because of the existence of the first and
second optical waveguides and the first and second optical
couplers. If such is the case, the signal light sent into the
output optical coupler from the third intermediate optical path and
the one sent into the output optical coupler from the reference
optical path can be made equivalent to ensure perfect switching by
arranging a second Mach-Zehnder interferometer having a
configuration same as that of the first Mach-Zehnder interferometer
on the reference optical path side. When the optical switch is used
as a router, it is not necessary to introduce control light into
the second Mach-Zehnder interferometer from the second control
light input waveguide.
High speed optical switching can be realized and will not be
affected by the carrier lifetime if the first and second optical
waveguides are active semiconductor waveguides that are so biased
as to establish an equilibrium between the gain and the loss
relative to low-power light having the wavelength of control light.
If semiconductor waveguides are used for the first and second
optical waveguides of a device according to the present invention,
they are preferably active transparent optical waveguides to
minimize the loss and maximize the nonlinearity because the optical
waveguide section of the device is apt to be rather long. If the
third and fourth optical waveguides are employed, they are
preferably also active transparent optical waveguides as the first
and second ones.
If an optical control type optical switch according to the present
invention is monolithically formed on a semiconductor substrate,
the Mach-Zehnder interferometers can be symmetrically arranged with
ease. If such is the case, since the number of contact points of
the optical waveguides is reduced and the components are uniformly
affected by temperature change, the device can enjoy enhanced
stability and reliability to say nothing of a reduced connection
loss and the possibility of downsizing and lowering the
manufacturing cost.
If the first and second optical waveguides show small discrepancy
in the symmetrical arrangement, an imperfect separation of signal
light and control light may result in the second optical coupler,
although such discrepancy can be compensated by inserting means for
shifting the phase of one of the optical waveguides relative to
that of the other one. This description holds true also for the
second Mach-Zehnder interferometer constituted by the third and
fourth optical waveguides and for the third Mach-Zehnder
interferometer constituted by the first Mach-Zehnder interferometer
and the reference optical path (or the second Mach-Zehnder
interferometer). Particularly if the signal branching means is
designed to realize symmetrical Y branching and hence the branched
two optical paths are perfectly symmetrical, the branching ratio of
the first output optical path to the second one is 1:1 in the
absence of control light. This means that the phase of the output
of either one of the optical paths has to be biased by .pi./2
relative to the phase of the output of the other one from the very
beginning to allow only one of the signal light output optical
paths to produce its output. Such phase regulating means may be
realized by arranging a phase modulator on the optical waveguides
or introducing biasing light from the second control light input
waveguide.
The third aspect of the present invention as described hereinafter
is essentially characterized by the use of a semiconductor optical
waveguide for a wavelength conversion device in order to enhance
the conversion efficiency, the semiconductor optical waveguide
comprising a semiconductor layer having a resonant wavelength of
intra-band absorption arranged within the gain bandwidth of the
traveling wave type semiconductor laser amplifier of the
device.
More specifically, a wavelength conversion device according to the
third aspect of the present invention comprises a traveling wave
type semiconductor laser amplifier having a semiconductor optical
waveguide formed by sandwiching an active layer between a pair of
clad layers, wherein light having an angular frequency different
from the light introduced into the semiconductor optical waveguide
is generated by four wave mixing in the semiconductor optical
waveguide, wherein the semiconductor optical waveguide has a
semiconductor layer having an intra-band absorption resonant
wavelength arranged within the gain band of the traveling wave type
semiconductor laser amplifier.
The semiconductor layer may be the active layer itself, part of the
clad layers or a layer independent from the active layer and the
clad layers so long as it covers part of the power distribution
zone of light guided through the semiconductor optical
waveguide.
For the purpose of the present invention, intra-band absorption
specifically refers to inter-valence band absorption from a heavy
or light hole band to a spin-orbit separation band and
inter-subband absorption of quantum well.
Four wave mixing is a nonlinear process of mixing three input
lights in a nonlinear medium to produce a fourth output light. The
three input light waves are two exciting lights and a signal light,
of which the two exciting lights may be same. If such is the case,
the number of input lights is in fact only two.
Assume here that a first light having an angular frequency of
.omega..sub.1 and a second light having an angular frequency of
.omega..sub.2 =.omega..sub.1 -.OMEGA. (where .OMEGA. is not equal
to zero) found within the gain wavelength band of the traveling
wave type semiconductor laser amplifier are introduced into the
device. Then, a third light having an angular frequency of
.omega..sub.3 =.omega..sub.1 +.OMEGA. is generated and produced as
a result of four wave mixing in the semiconductor optical
waveguide.
The nonlinear susceptibility of a wavelength conversion device
according to the third aspect of the present invention attributable
to inter-subband transition of the conduction band and the valence
band and/or intra-band transition (intra-band absorption) such as
inter-valence band transition is greater than the nonlinear
susceptibility attributable to inter-band transition.
If the wavelength of incident light agree with the resonant
wavelength of intra-band absorption, the carrier energy
distribution within the band changes remarkably as a result of
intra-band resonant absorption to consequently change the
absorption coefficient and the refractive index so that a
conversion efficiency higher than the conversion efficiency
(=.alpha..times.complex coupling coefficient, where .alpha. is a
constant) given rise to by carrier heating when no intra-band
resonant absorption takes place can be obtained.
Thus, a wavelength conversion device according to the present
invention and comprising a semiconductor optical waveguide having
semiconductor layer with a resonant wavelength of intra-band
absorption arranged within the gain bandwidth of the traveling wave
type semiconductor laser amplifier of the device can provide a
conversion efficiency higher than that of conventional wavelength
conversion devices without intra-band resonant absorption.
With a wavelength conversion device according to the present
invention, carriers excited to a high energy level return to the
original energy level by relaxation as described below.
The process of relaxation involves transition to a lower energy
band through collisions with phonons, relaxation to the Fermi
distribution through collisions of carriers and relaxation of the
carrier temperature to the lattice temperature through collisions
with phonons. These modes of relaxation are basically identical
with the modes of relaxation from spectral hole burning and from
carrier heating. The duration of the process of relaxation is in
fact a function of relaxation from carrier heating that proceeds
most slowly of all the above listed modes but takes only as short
as several hundred femtoseconds to several picoseconds. Therefore,
even if exciting light and signal light show a large difference in
wavelength, the fall of the conversion efficiency remains very
small over a large bandwidth.
Thus, according to the third aspect of the invention, there is
provided a wavelength conversion device that can carry out
wavelength conversion highly efficiently over a large
bandwidth.
A semiconductor optical waveguide device according to the fourth
aspect of the present invention comprises an optical waveguide
constituted by a second semiconductor optical waveguide layer made
of a material having a bandgap sufficiently greater than that of a
first semiconductor optical waveguide layer including an active
layer or that of the active layer itself, means for injecting an
electric current into the active layer and means for electrically
controlling the resonant wavelength of intra-band absorption of the
second semiconductor optical waveguide layer, characterized in that
the resonant wavelength of intra-band absorption of the second
semiconductor optical waveguide layer is arranged within the
stimulated emission gain wavelength bandwidth given rise to by
electric current injection into the active layer.
In the semiconductor optical waveguide device as described above,
the means for controlling the resonant wavelength of intra-band
absorption of the second semiconductor optical waveguide layer may
be operated as means for applying an electric field to the second
semiconductor optical waveguide layer.
A material having a bandgap more than twice, preferably three
times, as large as that of the active layer of the first
semiconductor optical waveguide layer may be used for the second
semiconductor optical waveguide layer. Materials that can be used
for the first semiconductor optical waveguide layer to suitably
meet the above requirement include InP and InGaAsP, while materials
that can be used for the second semiconductor optical waveguide
layer to meet the above requirement include InGaN, GaN and AlN. The
second semiconductor optical waveguide layer may have a quantum
well structure. For the purpose of the present invention,
intra-band resonant absorption specifically refers to inter-subband
absorption of quantum well, and inter-valence band absorption.
A semiconductor optical waveguide device according to the present
invention can be realized in a number of different modes as
described below to provide different functional features. In a mode
of realization, the second semiconductor optical waveguide layer is
arranged such that the spectrum of intra-band resonant absorption
may be varied by applying an electric field. If a semiconductor
optical waveguide device according to the present invention
involves inter-subband absorption, an asymmetric well structure may
be used in order to realize a large change in the inter-subband
transition energy transition by applying an electric field.
The second semiconductor optical waveguide lay may be at least
partly doped with one or more than one impurities. The first and
second semiconductor optical waveguide layers may be placed close
to each other in a layered arrangement or, alternatively, serially
connected to each other. Still alternatively, the optical waveguide
may comprise a portion constituted by only either one of the first
and second semiconductor optical waveguide layers and a portion
where the two layers are placed close to each other in a layered
arrangement. Still alternatively, a third semiconductor optical
waveguide layer may be arranged between the first and second
semiconductor optical waveguide layers.
A single or more than one intra-band absorption resonant
wavelengths of the second semiconductor optical waveguide layer may
be found within the gain bandwidth of the active layer. The second
semiconductor optical waveguide layer may be divided into a
plurality of zones along the optical waveguide that are
respectively provided with independent means for controlling the
intra-band absorption resonant wavelength by applying an electric
field to the second optical waveguide layer. If such is the case,
the zones may have different respective intra-band absorption
resonant wavelengths. Of the plural zones, some may have a
relatively large well width, whereas the rest may have a relatively
small well width.
The optical waveguide may be provided with means for preventing
multiple reflections of traveling light from taking place along the
traveling direction. If such is the case, a semiconductor optical
waveguide device according to the present invention is a traveling
wave type optical waveguide device. Such means specifically refers
to formation of an anti-reflection film on each of the input and
output facets of the device, use of a window structure, monolithic
integration of the optical waveguide with other devices and angled
facets relative to the optical waveguide.
If a semiconductor optical waveguide device is used as a resonance
type device such as a semiconductor laser, it may be provided with
optical feedback means in order to give rise to resonance of light
having a specific wavelength. For the optical feedback means, a
diffraction grating, a cleaved surface or an etched surface may be
used.
In a semiconductor optical waveguide device according to the fourth
aspect of the present invention, a stimulated emission gain may be
produced over a wide wavelength band near the band edge energy
level by injecting an electric current into the active layer of the
first semiconductor optical waveguide layer. Since the second
semiconductor optical waveguide layer is constituted by a
semiconductor layer having a bandgap sufficiently greater than that
of the active layer, it does not give rise to any inter-band
absorption but produces intra-band absorption for light having a
wavelength corresponding to the gain band wavelength of the active
layer. Since the magnitude of intra-band absorption depends on the
carrier density of the second semiconductor optical waveguide
layer, a desirable level may be obtained by selecting an
appropriate concentration for the impurities to be doped. The
intra-band resonant absorption spectrum has a width as small as
tens of several meV at most, which is by far smaller than the width
of the interband absorption spectrum and that of the gain spectrum
of the active layer.
Since light being transmitted through the optical waveguide is
sensitive to both any gain of the active layer and absorption in
the second semiconductor optical waveguide layer, gain holes may be
produced in the gain spectrum obtained as a net result involving
the gain of intra-band absorption. If a plurality of intra-band
absorption resonant wavelengths are provided within the gain band
of the active layer, an active wavelength filter having a variety
of transmission spectra may be formed by appropriately arranging
overlapped and/or isolated gain holes. Since the active layer has a
gain, a gain is produced in the transmission wavelength range
whereas a loss is provided in the cut-off wavelength range.
As an electric field is applied to the second semiconductor optical
waveguide layer, the gain holes change their positions and sizes.
The transmissivity may be controlled in a more sophisticated way if
the second semiconductor optical waveguide layer and the means for
applying a voltage thereto are divided into a plurality of zones.
Any change in the intra-band absorption spectrum instantaneously
follows up the change in the electric field, be it caused by
inter-subband absorption of quantum well or by inter-valence band
absorption.
If the device is of the traveling wave type provided with means for
preventing multiple reflections of light along the traveling
direction, the high speed optical pulse having a transmission
wavelength is transmitted without entailing any strained or divided
pulse waveform attributable to multiple reflections. When an
optical pulse having a cut-off wavelength is introduced, carriers
excited to a high energy level by intra-band absorption is restored
to the original energy level by relaxation within a very short
period of several picoseconds even if they are not drawn out to the
outside by an electric field, so that consequently no pattern
effect is produced by saturated absorption.
Because the absorption coefficient for a specific wavelength is
changed by applied an electric field to the second semiconductor
optical waveguide layer, a semiconductor optical waveguide device
according to the present invention can be applied to an optical
intensity modulator.
Generally speaking, if an absorption spectrum is modified
remarkably by applying an electric field, the refractive index of
the wavelength close to that also changes significantly. In view of
this phenomenon, it will be understood that a semiconductor optical
waveguide device according to the present invention can be applied
to a device that utilizes changes in the refractive index. Since an
absorption spectrum has a spectral width smaller than that of
inter-band absorption, a phase modulator showing a small change of
absorption and a large change of refractive index can be realized
by using a wavelength located just outside that of the absorption
peak.
Additionally, by appropriately selecting a wavelength and a voltage
to be applied, the ratio (.alpha. parameter) of a change in the
absorption coefficient to a change in the refractive index produced
by the power of control light and an electric field can be modified
significantly.
Still additionally, if the present invention is applied to an
optical control type optical switch that utilizes cross phase
modulation within a traveling wave type optical amplifier or a
wavelength conversion device that utilizes four wave mixing and
gain saturation within a traveling wave type optical amplifier, the
absorption coefficient, the conversion efficiency and other
operating parameters of such a device can be controlled by way of
external voltage terminals.
Still additionally, a photodetector having a high speed wavelength
tuning capability can be realized by providing additional means to
draw carriers with a high energy level produced by intra-band
absorption to the outside by resonance tunneling or some other
technique.
Still additionally, if the present invention is applied to a
resonance type optical waveguide device by using optical feedback
means, it may be used for a resonance type wavelength filter, a
resonance type optical control switching device or a
multifunctional light source. For instance, if the second
semiconductor optical waveguide layer is arranged within the
resonator, it may be used for a tunable wavelength laser that can
quickly change the wavelength, a mode-locked laser wherein the
second semiconductor optical waveguide layer is constituted by a
saturable absorber whose performance can be controlled by a voltage
or a laser light source that can generate short pulses by using a
loss switch. If the second semiconductor optical waveguide layer is
arranged outside the resonator, a light source comprising
integrated intensity modulators or integrated phase modulators may
be realized.
If a material having a large bandgap generously exceeding the twice
of the bandgap of the active layer is used for the second
semiconductor optical waveguide layer, changes in the carrier
density of the second semiconductor optical waveguide layer
attributable to absorption of light being transmitted through the
optical waveguide by two-photon absorption can be prevented from
taking place so that the slow fluctuation of the response
restricted by the carrier lifetime (2 several hundred picoseconds)
is suppressed and fast and stable operation is realized.
Generally, multiple photon absorption that absorb N photons can be
prevented from taking place by reducing the wavelength of the
interband absorption in the second semiconductor optical waveguide
layer to a level sufficiently shorter than 1/N of the optical
wavelength that is being used.
In short, a semiconductor optical waveguide device according to the
present invention can be used for a light source, a photodetector
device, an optical waveguide device for modifying the optical
spectrum, the intensity of light or the phase of light or a device
having complex functional features.
To summarize up, the semiconductor optical waveguide device
comprises a stripe-shaped semiconductor optical waveguide, part of
the semiconductor optical waveguide being an active layer producing
gain by electric current injection, and another part of the
semiconductor optical waveguide being an intra-band resonant
absorption layer in which an intra-band absorption resonant
wavelength is arranged within the gain band of the active layer,
and means for injecting electric current into the active layer.
Additional objects and advantages of the present invention will be
set forth in the description which follows, and in part will be
obvious from the description, or may be learned by practice of the
present invention. The objects and advantages of the present
invention may be realized and obtained by means of the
instrumentalities and combinations particularly pointed out in the
appended claims.
BRIEF DESCRIPTION OF THE DRAWINGS
The accompanying drawings, which are incorporated in and constitute
a part of the specification, illustrate presently preferred
embodiments of the present invention and, together with the general
description given above and the detailed description of the
preferred embodiments given below, serve to explain the principles
of the present invention in which:
FIGS. 1A through 1C are graphs showing the change with time of the
phase of a transmitted probe light pulse after the transmission of
exciting light pulse;
FIG. 2 is a schematic cross sectional view of a conventional
nonlinear directional coupler;
FIG. 3 is a graph showing the output performance of a conventional
nonlinear direction coupler;
FIG. 4 is a schematic view of a wavelength conversion system using
a conventional wavelength conversion device;
FIG. 5 is a graph showing the .OMEGA. dependency of the wavelength
conversion efficiency .eta. of a tensile strained InGaAs/InGaAsP
(MQW) conventional traveling wave type semiconductor laser
amplifier;
FIG. 6 is a schematic illustration of wavelength conversion nodes
using conventional traveling wave type semiconductor laser
amplifiers and a tunable wavelength filter;
FIG. 7 is a schematic illustration of an optical control type
optical switch realized by using a traveling wave type
semiconductor laser amplifier in a first embodiment according to
the present invention, showing its configuration;
FIG. 8 is a schematic perspective view of a traveling wave type
semiconductor laser amplifier according to a first embodiment or a
wavelength conversion device according to a seventh embodiment of
the present invention;
FIG. 9 is a graph showing the composition X dependency of the band
gap Eg and the spin orbit split-off energy .DELTA..sub.o of
HgCdTe;
FIG. 10 is a schematic illustration of intervalence band absorption
and its relaxation process of HgCdTe;
FIG. 11 is a schematic perspective view of an optical control type
optical switch according to a second embodiment of the present
invention;
FIG. 12 is a schematic cross sectional view of the optical control
type optical switch of FIG. 11, showing its optical waveguide
section;
FIG. 13 is a schematic illustration of the conduction band of a
principal portion of the active optical waveguide of the optical
control type optical switch of FIG. 11, showing its structure;
FIG. 14 is a schematic view of a Mach-Zehnder interferometer type
optical switch according to a third embodiment of the present
invention, showing its configuration;
FIG. 15 is a schematic cross sectional view of the optical
waveguide section of the Mach-Zehnder interferometer type optical
switch of FIG. 14;
FIG. 16 is a schematic illustration of the conduction band of the
nonlinear optical waveguide layer of FIG. 15, showing its
structure;
FIGS. 17A through 17F are schematic cross sectional views of a
Mach-Zehnder interferometer type optical switch in different
manufacturing steps according to the third embodiment of the
present invention;
FIG. 18 is a schematic illustration of an optical control type
optical switch according to a fourth embodiment of the present
invention, showing its configuration;
FIGS. 19A and 19B are schematic cross sectional views of the
optical control type optical switch of FIG. 18 taken along the
direction of waveguide and the direction perpendicular to that of
waveguide respectively;
FIG. 20 is a pulse timing chart of the operation of the optical
control type optical switch of FIG. 18;
FIG. 21 is a schematic illustration of an optical demultiplexer
realized by applying the optical control type optical switch of
FIG. 18, showing its configuration;
FIG. 22 is a graph showing the input-output relationship of an
analog optical modulator realized by applying the optical control
type optical switch of FIG. 18;
FIG. 23 is a schematic perspective view of a wavelength conversion
device according to a fifth embodiment of the present
invention;
FIG. 24 is a schematic illustration of the conduction band of the
strained quantum well active layer of the optical waveguide of the
wavelength conversion device of FIG. 23, showing its structure;
FIG. 25 is a schematic illustration showing how the state of
carriers of the wavelength conversion device of FIG. 23
changes;
FIG. 26 is a graph showing the relative wavelength conversion
efficiency of the wavelength conversion device of FIG. 23 in
comparison with that of a conventional wavelength conversion
device;
FIG. 27 is a schematic perspective view of a wavelength conversion
device according to a sixth embodiment or a tunable wavelength
filter according to an eighth embodiment of the present
invention;
FIGS. 28A and 28B are graphs showing transmission spectrums of the
tunable wavelength filter of FIG. 27;
FIG. 29 is a schematic illustration of a wavelength conversion node
using a tunable wavelength filter as illustrated in FIG. 27;
FIG. 30 is a schematic cross sectional view of a tunable wavelength
DFB laser according to a ninth embodiment of the present invention,
showing its optical waveguide;
FIG. 31 is a graph showing absorption spectrums (dotted lines) and
the change in the refractive index (solid line) of the tunable
wavelength DFB laser of FIG. 30 obtained by applying a variable
voltage;
FIGS. 32A and 32B are schematic cross sectional views of devices
obtained by modifying the semiconductor optical waveguide device of
FIG. 30; and
FIG. 33 is a table showing various parameters concerning the
optical switch according to the present invention, showing expected
advantages.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Now, the present invention will be described further by referring
to the accompanying drawings that illustrate preferred embodiments
of the present invention.
(1st Embodiment)
FIG. 7 is a schematic illustration of a Mach-Zehnder interferometer
type optical switch realized by using a traveling wave type
semiconductor laser amplifier according to the first embodiment of
the present invention, showing its configuration.
As shown, components arranged between a polarization conserving
fiber type 1:1 input optical coupler 1 and a polarization
conserving fiber type output optical coupler 2 constitutes a
Mach-Zehnder interferometer. A first fiber type polarization
coupler 3, a semiconductor optical waveguide device (traveling wave
type semiconductor laser amplifier) 20 according to the present
invention and a second fiber type polarization coupler (splitter) 4
are arranged on the first branch of the interferometer while an
LiNbO.sub.3 intensity modulator 5 and a phase modulator 6 are
arranged on the second branch of the interferometer.
Signal light is pulse light having a linearly polarized wave with a
wavelength of 1.3 .mu.m and entered through an input terminal 11 of
the input optical coupler 1 and made to go out through output
terminals 13, 14 of the output optical coupler 2. Control light has
a wavelength substantially equal to that of signal light and has a
polarization orthogonal to that of the signal light. Control light
is entered via the first polarization coupler 3 and made to go out
via the second polarization coupler 4. The peak power of signal
light is less than 1 mW, whereas that of control light is about 200
mW. Their pulse widths are respective 2 ps and 5 ps.
FIG. 8 schematically illustrates the traveling wave type
semiconductor laser amplifier 20.
In FIG. 8, reference numeral 21 denotes a p-type CdTe substrate
that also operates as a p-type clad layer. A semiconductor optical
waveguide 22 is formed on the p-type CdTe substrate 21.
More specifically, the semiconductor optical waveguide 22 is formed
by sequentially laying the p-type clad layer (p-type CdTe
substrate) 21, an Hg.sub.0.3 Cd.sub.0.7 Te active waveguide layer
24, a mesa-shaped n-type CdTe clad layer 25 to produce a multilayer
structure and the waveguide is defined by the stripe-shape of the
n-type CdTe clad layer 25. An anti-reflection film 23 is arranged
on each of the input and output facets of the semiconductor optical
waveguide 22. An n-side electrode 26 is arranged on the mesa
section of the n-type CdTe clad layer 25, whereas a p-side
electrode 27 is arranged on the p-type CdTe substrate 21. The
active waveguide layer 22 is biased by injecting an electric
current through the electrodes 26 and 27 to show an active
transparent state, where the loss and the gain are balanced.
The intensity modulator 5 regulates the intensity of the light
introduced from the second branch so as to make it equal to that of
the light introduced from the first branch. The phase modulator 6
regulates the phase of the second branch so as to make all signal
output come out of the first output terminal 13 if there is not
control light.
If a control light pulse is introduced in synchronism with the
signal light pulse under this condition, the phase of the signal
light is modified by the nonlinear optical effect of the Hg.sub.0.3
Cd.sub.0.7 Te active waveguide layer 24 that is biased to an active
transparent state. This process of modification will be described
below. If the intensity of the control light pulse is so regulated
as to achieve a phase modification exactly equal to .pi. at the
output end, the signal output of the output optical coupler 2 is
switched to the output terminal 14. This operation is carried out
within 2 picoseconds as described earlier by referring to the prior
art.
FIG. 9 is a graph showing the composition dependency of the band
gap Eg and the spin orbit splitting .DELTA..sub.o of Hg.sub.1-x
Cd.sub.5 Te. It will be seen that Eg.apprxeq..DELTA..sub.o for
Hg.sub.0.3 Cd.sub.0.7 Te (where x=0.7) with a resonant wavelength
of about 1.3 .mu.m. When guided light having this wavelength is
introduced into the Hg.sub.0.3 Cd.sub.0.7 Te active waveguide layer
24, holes are excited to the spin orbit split-off band by
inter-valence band resonant absorption. Note that the net
inter-band transition is suppressed because the active waveguide
layer is held to a transparent state.
In the above described system, the effective mass of each valence
electron at or near the .GAMMA. point of the Brillouin zone is
about 0.4 m.sub.o and there is only a small difference between any
two effective masses (differently stated, the dispersion curve is
parallel and joint-density-of-state is large) to show a large
absorption coefficient. Consequently, hot holes are generated
efficiently by strongly exciting light. As the energy distribution
of holes changes, the refractive index also changes. Since the
number of holes in the heavy and light holes bands decreases under
this condition, the gain also decreases instantaneously.
As shown in FIG. 10, excited hot holes lose their energy in a very
short period of time as a result of inter-carrier collisions and
collisions with phonons. The time required for intra-band
relaxation of hot carriers is less than 0.1 picoseconds. Other
carriers are also heated by the energy discharged as a result of
relaxation. Then, any heated carriers also lose gradually the
energy they have as they collide with phonons to restore the
original state within 1 picosecond. In other words, as an exciting
pulse is applied, there immediately occurs a large change in the
refractive index but the refractive index and the transmission
coefficient returns to the original values in about 1 picosecond
after the removal of the exciting pulse.
Strictly speaking, the carrier density is also changed slightly as
a result of incidental two-photon absorption. However, since the
nonlinearity due to inter-valence band absorption is enhanced as a
result of resonance, a switching operation can be carried out with
a power level of exciting light that is by far lower than that of
its counterpart of a conventional active transparent optical
waveguide so that the influence of two-photon absorption that is
proportional to the square of the power can be minimized.
Additionally, since electrons and holes are already present at a
high density in the active layer 22, the generation of carriers due
to impact ionization caused by hot holes is suppressed. Thus, any
undesirable changes in the performance of the device accompanied by
an increased time constant that can be give rise to by excessively
generated and accumulated carriers can also be suppressed even if
the switching operation is repeated at an enhanced frequency of
several hundred Gb/s.
As described above, with the first embodiment, since the
nonlinearity due to intra-band absorption is enhanced by resonance,
a switching operation can be carried out with a power level of
exciting light that is by far lower than that of its counterpart of
a conventional active transparent optical waveguide. Additionally,
since the influence of two-photon absorption that excites carriers
beyond the bandgap can be minimized, a high speed switching
operation that is not restricted by the carrier lifetime can be
realized.
(2nd Embodiment)
FIG. 11 is a schematic perspective view of an optical control type
optical switch according to the second embodiment of the present
invention, and FIG. 12 is a schematic cross sectional view of the
optical control type optical switch of FIG. 11, showing its optical
waveguide section.
The optical switch is formed on an n-InP substrate 31 and a
directional coupler is formed at the center thereof by a pair of
mesa-shaped active optical waveguides 32a, 32b. The active optical
waveguides 32a, 32b are connected at the opposite ends thereof near
the ends of the device to mesa-shaped passive optical waveguides
33a, 33b, 33c, 34d respectively. Each of the active optical
waveguides 32a, 32b comprises an undoped InGaAsP passive waveguide
layer 34, a thin undoped InP etch-stop layer 35, an InGaAs/strained
AlAs quantum well layer 36, an InGaAsP waveguide layer 37, a p-type
InP clad layer 38 and a p-type InGaAsP ohmic contact layer 39 laid
sequentially in the cited order on the substrate 31 to form a
multilayer structure. Under the passive optical waveguides 33a
through 33d, the undoped InGaAsP passive waveguide layer 34 is
sandwiched between the substrate 31 and a semi-insulated InP layer
40.
The active optical waveguides 32a, 32b carriers thereon respective
ohmic electrodes 41a, 41b, while another ohmic electrode 42 is
formed under the substrate 31. The electrodes 41a, 41b are
connected to respective pads (not shown) on an insulation film and
also to external circuits by bonding. The input and output facets
of the optical switch is provided with an anti-reflection coat 43.
The entire device is arranged on a Cu mount by way of the lower
electrode 42, said Cu mount also operating as a heat sink and a
ground.
FIG. 13 is a schematic illustration of the conduction band of a
principal portion of the strained quantum well active layer 36 of
either one of the active optical waveguides 32a, 32b. The quantum
well layer 36 is realized by regularly arranging twenty five (25)
unit structures, each comprising a thin InGaAs well layer 44 and a
thin tensile strained AlAs barrier layer 45, in such a way that
each InGaAs layer 44 is sandwiched between a pair of AlAs barrier
layers 45, 45. A pair of subbands 46, 47 are arranged within each
well layer. Since the barrier layer is thin, the subbands 46, 47 of
each well are coupled by the tunneling effect to produce a
miniband. Its inter-subband transition energy is about 0.8 eV
(resonant wavelength of 1.55 .mu.m) relative to TM mode light. The
fact that such a large interval can be formed between a pair of
subbands is described in J. H. Smet et al., Appl. Phys. Lett.,
vol.64, pp. 986-987, 1994.
The active optical waveguides 32a, 32b are so biased as to become
transparent relative to TM mode light having a wavelength of 1.55
.mu.m and have a small gain relative to TE mode light. Thus,
electrons are injected into the first subband 46 by means of
tunneling, whereas the second subband 47 is normally held
vacant.
A weak signal light pulse having a wavelength of 1.55 .mu.m is
introduced to the active transparent optical waveguide 32a in the
TE mode. If no exciting light pulse is present, the directional
coupler is in a completely coupled state and the signal light pulse
is transmitted to 33d. On the other hand, a strong exciting light
pulse having a wavelength of 1.55 .mu.m is introduced into the
other active transparent optical waveguide 32b in the TM mode. If
there is a strong exciting light pulse, the refractive index of the
active transparent optical waveguides 32a, 32b is modified by the
Kerr effect to switch the destination of signal light to 33c.
Signal light can be separated from signal light typically by a
polarization coupler.
The second embodiment operates substantially same as the first
embodiment. If strong exciting wave having a wavelength of 1.55
.mu.m is introduced into either one of the active transparent
optical waveguides 32a, 32b, excitation takes place from the first
subband 46 to the second subband 47 as a result of inter-subband
resonant absorption, although net inter-band transition is
suppressed. Since inter-subband transition is tolerative to light
in the TM mode and the nonlinearity of inter-subband transition is
generally large, there arises a large change in the refractive
index (Kerr effect).
Electrons excited in the second subband 47 loses their energy in a
short period of time as a result of inter-carrier collisions and
collisions with phonons. The time required for relaxation for
inter-subband and intra-band electrons is less than 0.1
picoseconds. Other carriers warmed by this relaxation gradually
lose their energy as a result of collisions with phonons. If
electrons are scattered to the L and X points in the process of
relaxation, about 1 picosecond will have to be spent for them to
return to the .GAMMA. point to make the recovery time slightly
longer than that of the first embodiment, although they restore the
original state in several picoseconds after the removal of the
exciting pulse in any case. In short, the refractive index
experiences a remarkable change instantaneously when an exciting
light pulse is introduced but recovers its original level as soon
as the exciting light pulse is gone.
With this second embodiment again, the influence of two-photon
absorption is minimized to reduce the power required for exciting
light. Additionally, since electrons and holes are already present
at a high density in the active transparent optical waveguide layer
32, the generation of carriers due to impact ionization caused by
hot electrons is suppressed. Thus, any undesirable changes in the
performance of the device accompanied by an increased time constant
that can be give rise to by excessively generated and accumulated
carriers can also be suppressed even if the switching operation is
repeated at an enhanced frequency of several hundred Gb/s.
As described above, with the second embodiment, a switching
operation can be carried out with a power level of exciting light
that is by far lower than that of its counterpart of a conventional
active transparent optical waveguide.
It should be noted that the present invention is not limited to the
above described embodiments and they may be modified or changed in
a number of different ways. For example, they may be operated with
a wavelength other than the one cited above by artificially
changing the bandgap, the inter-valence band absorption energy and
the inter-subband transition energy by means of a strained
superlattice and/or a strained quantum well. The nonlinearity may
be made to become more remarkable by appropriately modifying
various parameters of the material of the active optical waveguide
such as effective mass by means-of strain. Additionally, the
material of the active optical waveguide and the overall
configuration of the optical switch are not limited to those
described above by referring to the above embodiments. For
instance, the use of an intensity modulator and a phase modulator
is not necessarily indispensable and an active transparent optical
waveguide may be arranged in each of the branches of the
Mach-Zehnder interferometer of the first embodiment. The optical
switch, the optical coupler and other related devices may be
monolithically formed on a semiconductor substrate.
(3rd Embodiment)
FIG. 14 is a schematic view of a Mach-Zehnder interferometer type
optical switch according to the third embodiment of the present
invention.
The Mach-Zehnder interferometer comprises nonlinear waveguide
sections 1002a, 1002b, 3 dB couplers 1003a, 1003b and polarization
couplers 1004a through 1004d. The optical input/output ports of the
interferometer include signal light input ports 1005a, 1005b,
signal light output ports 1006a, 1006b, a control light input port
1007a, a control light output port 1008a and dummy input and output
ports 1007b and 1008b for maintaining the symmetry of the
interferometer. These ports are connected to the device by way of
respective polarization maintaining fibers 1009.
The polarized waves of signal light and control light introduced
into the respective polarization couplers 1004a and 1004b are so
arranged that the signal light and the control light are
transmitted through the respective nonlinear waveguide sections
1002a and 1002b in the TE mode and the TM mode respectively. The
wavelengths of signal light and control light are both equal or
close to 1.55 .mu.m. Signal light introduced through the input port
1005a is divided into two branches by the 3 dB coupler 1003a to a
ratio of 1:1, one of which is led to the 3 dB coupler 1003b by way
of the polarization coupler 1004a, the nonlinear waveguide section
1002a and the polarization coupler 1004c while the other is also
led to the 3 dB coupler 1003b by way of the polarization coupler
1004b, the nonlinear waveguide section 1002b and the polarization
coupler 1004d. If no control light is present, the signal light
experiences interference in the 3 dB coupler 1003b and sent out to
the output port 1006b. On the other hand, control light is
introduced through the control light input port 1007a and combined
with the signal light in the polarization coupler 1004a to shift
the phase of the signal light by .pi. in the nonlinear waveguide
section 1002a before it is separated from the signal light and fed
to the control light output port 1008a. As the phase of signal
light is shifted by .pi. by control light in the nonlinear
waveguide section 1002a, the signal light output is switched to the
output port 1006a as a result of interference in the 3 dB coupler
1003b.
It may be understood from the above description that the
configuration and the operation of this embodiment are basically
same as those of any conventional Mach-Zehnder interferometer
nonlinear optical switch, although the third embodiment is
characterized by the arrangement of nonlinear waveguide sections
1002a, 1002b and their functions and effects. FIG. 15 is a
schematic cross sectional view of the Mach-Zehnder interferometer
type optical switch of FIG. 14 taken along a plane perpendicular to
the waveguide.
Each of the nonlinear waveguide sections 1002a, 1002b comprises as
principal components an InGaN/GaN/AlN multiple quantum well
nonlinear waveguide layer 1012 formed on a (0001) sapphire
substrate 1001 with an AlN buffer layer 1011 arranged therebetween,
a thin upper AlGaInN buffer layer 1013, an n-type InGaAsP layer
(with a PL wavelength of 1.15 .mu.m) 1014 integrally formed on the
buffer layer by direct bonding, a mesa-shaped undoped InGaAsP
active waveguide layer (with a PL wavelength of 1.55 .mu.m) 1015
formed thereon, a p-type InP clad layer 1016 and a p-type InGaAs
contact layer 1017. The nonlinear waveguide layer 1012 and the
buffer layer 1013 are arranged in so many layers with the c-axis
agreeing with the normal. The layers from the n-type InGaAsP layer
1014 up to the p-type InGaAs contact layer 1017 are arranged in so
many layers along the direction of <001>.
Light is guided mainly through the n-type InGaAsP layer 1014 and
the active waveguide layer 1015, although it also permeates into
nonlinear waveguide layer 1012 a nd the p-type clad layer 1016 to a
large proportion to produce a unitary optical waveguide. The
optical nonlinearity of the nonlinear waveguide section 1002 is
mainly realized by the InGaN/GaN/AlN multiple quantum well
nonlinear waveguide layer 1012 and the InGaAsP active waveguide
layer 1015 operates mainly as gain producing means for compensating
the produced waveguide loss.
On the n-type InGaAsP layer 1014, a pair of plateau-like InP
contact layers 1018a, 1018b are formed outside of the two waveguide
sections 1002a, 1002b with a distance of several .mu. m arranged
therebetween for separating them. The gaps separating the optical
waveguide sections 1002a, 1002b and the n-type contact layers
1018a, 1018b are mostly filled with polyimide 1019. A p-type ohmic
electrode 1020 is formed on the p-type contact layer 1017, while an
n-type ohmic electrode 1021 is formed on each of the n-type contact
layers 1018a, 1018b. Metal pads 1022a, 1022b are formed on the
ohmic electrodes 1020, 1021 and part of the polyimide layer and
metal wires 1023a, 1023b are bonded thereto in order to establish
electric connection with the outside.
An electric current fed from the outside flows sequentially through
the wire 1023a, the pad 1022a, the p-type ohmic electrode 1020, the
p-type contact layer 1017, the p-type clad layer 1016, the active
waveguide layer 1015, the n-type InGaAsP layer 1014, the n-type
contact layer 1018, the n-type ohmic electrode 1021, the pad 1022b
and the wire 1023b. As the electric current is injected, a gain is
produced in the active waveguide layer 1015 to compensate the loss
produced in the nonlinear waveguide layer 1012, the p-type clad
layer 1016 and some other layers.
FIG. 16 schematically illustrates the structure of the conduction
band of the nonlinear waveguide layer 1012. The nonlinear waveguide
layer 1012 has a multiple quantum well structure realized by
arranging ten (10) quantum wells, each comprising an undoped
In.sub.0.25 Ga.sub.0.75 N well layer 1031 having a thickness of
1.06 nm, an undoped AlN barrier layer 1032 having a thickness of
about 2 nm and a pair of n-type GaN intermediate layers 1033a,
1033b formed therebetween and having a thickness of 0.52 nm. Each
quantum well has three subbands, of which the first one is held to
an energy level lower than that of the bottom of the conduction
band of the intermediate layer 1033, whereas the second and third
subbands are held to an energy level higher than that of the bottom
of the conduction band of the intermediate layer 1033. The resonant
wavelength between the first and second subbands is about 1.554
.mu.m and the resonant wavelength between the second and third
subbands is about 1.21 .mu.m. Almost all electrons fed from the
n-type GaN intermediate layer 1033 are introduced into the first
subband.
Since the first and second subbands of this embodiment show
respective effective masses that are slightly different from each
other, the wave number-energy dispersion curves of the two subbands
eventually lose parallelism and the absorption spectrums tend to be
slightly widened. However, the effect of such phenomena is not
significant because electrons located near the .GAMMA. point really
account for inter-subband transition. On the other hand, changes in
the effective mass caused by inter-subband transition raise the
degree of nonlinearity.
FIGS. 17A through 17F are schematic cross sectional views of the
optical switch of the third embodiment in different manufacturing
steps. A technique of direct bonding is used for the semiconductor
substrate here.
Referring firstly to FIG. 17A, an AnN buffer layer 1011, an
InGaN/GaN/AlN multiple quantum well nonlinear waveguide layer 1012
and an AlGaInN buffer layer 1013 are made to grow on a (0001)
sapphire substrate 1001 by an MBE technique using a nitrogen plasma
source.
Apart from the substrate, an n-type InGaAsP layer 1014, an InGaAsP
active waveguide layer 1015, a p-type InP clad layer 1016, a p-type
InGaAs contact layer 1017 are sequentially formed on a (001) InP
substrate 1040 by epitaxial growth using an MOCVD technique.
Then, an insulation film 1041 is put to the epitaxial wafer, which
is subsequently subjected to a patterning operation to etch out the
contact layer 1017 and most of the p-type InP clad layer 1016 by
means of a CH.sub.4 RIE technique. Thereafter, the remaining p-type
InP clad layer 1016 is etched by wet selective etching and then
further wet-etched on a time-control basis to remove most of the
InGaAsP active waveguide layer 1015. Consequently, a mesa region
1042 extending along the direction of <110> is formed as
shown in FIG. 17C.
Again an MOCVD growth is carried out to bury the mesa region 1042
with an n-type InP layer 1018 until a substantially flat surface is
produced there. After removing the insulation film, the epitaxial
substrate is bonded to a glass substrate 1044 with an adhesive
agent 1043 and the InP substrate 1040 is etched from the rear side,
using a hydrochloric acid type selective etching solution. When the
etching operation is stopped at the InGaAsP layer 1014, a structure
as shown in FIG. 17D, where only the epitaxial layer is bonded to
the glass substrate 1044, is left there. A flat surface of the
n-type InGaAsP layer 1014 is exposed on the structure.
The surface of the n-type InGaAsP layer 1014 is mirror-polished to
a surface coarseness of less than 50 nm and treated with acid,
followed by washing with water and drying. Similarly, the surface
of the flat AlGaInN buffer layer 1013 of the substrate, on which a
nitride has been subjected to epitaxial growth as shown in FIG.
17A, is also mirror-polished to a surface coarseness of less than
250 nm and treated with acid, followed by washing with water and
drying. Thereafter, the treated two surfaces of the substrates are
directly bonded together by applying, if appropriate, pressure of
several kg/cm.sup.2 or without applying pressure. Since indium (In)
is present on the both surfaces as a component element in a highly
mobile state and phosphor (P) of the InGaAsP layer 1014 is also
highly mobile through a vapor phase, the surfaces can be bonded
together with relative ease. This bonding process is conducted in a
clean room with a class 10 rating in order to prevent particles of
dust from adhering the surfaces.
As the substrates are bonded together, the adhesive agent 1043 is
removed and the glass substrate 1044 is separated therefrom before
heating them at 250.degree. to 500.degree. C. in a hydrogen
atmosphere. This heat-treatment may be conducted under appropriate
pressure. As a result of the heat-treatment, the bonding strength
is raised to provide a strongly bonded structure as shown in FIG.
17E. An oxide film may or may not be disposed on the bonded
interface.
Thereafter, the bonded structure is again subjected to a series of
operations of patterning, RIE mesa etching and wet etching to
produce 2.5 .mu. m wide mesas of the nonlinear waveguide section
1002 and the n-type contact layer 1018 as shown in FIG. 17F. A
route is secured for electricity to run through by leaving the
n-type InGaAsP layer 1014 there at this stage of operation.
Then, the surface is coated with polyimide, which is subsequently
cured, and the mesas are made to become exposed. Thereafter, the
ohmic electrodes and the metal pad are formed and the bonded
substrates are cut to a given size to produce a finished nonlinear
waveguide having a cross section as schematically shown in FIG.
15.
The sapphire substrate 1001 is mounted on a heat sink (not shown)
whose temperature can be controlled. The input/output surfaces of
the nonlinear waveguide sections 1002a, 1002b are coated with an
anti-reflection film to realize a low loss optical coupling.
The figure of merit of an optical switch is normally expressed in
terms of .vertline..chi..sup.(3) .vertline./(.alpha..tau.). In
other words, a high speed high efficiency optical switch that
consumes energy at a reduced rate has an enhanced nonlinearity, a
small absorption coefficient (.alpha.) and a short response time
(.tau.). From experience, it is known that the value is somewhere
around 100 esu.cndot.cm/s(.apprxeq.1.4.times.10.sup.-8 m.sup.3
/V.sup.2) at most (D. H. Auston et al., Appl. Opt., vol.26,
pp.211-234, 1987). While it may be possible to achieve an
exceptionally large figure of merit by low temperature anthracene
surface excitation, the above value cannot be superseded by a
remarkably higher value at room temperature. In short, no
practically feasible high speed high efficiency nonlinear optical
switch has been known to data.
The transition matrix element for interaction between light and
substance is expressed by the approximate equation below.
where H.sub.I is the interaction Hamiltonian, .vertline.u.sub.i
> and .vertline.u.sub.f > are periodical portions of a Bloch
function parallel to the well plane at the initial and final states
respectively and .vertline.f.sub.i > and .vertline.f.sub.f >
are envelop functions perpendicular to the well at the initial and
final states respectively.
As different orthogonal base functions are applicable to the
valence band and the conduction band for inter-band transition,
<u.sub.f .vertline.u.sub.i >.sub.cell =0 so that <u.sub.f
f.sub.f .vertline.H, .vertline.u.sub.i f.sub.i >=<u.sub.f
.vertline.H.sub.I .vertline.u.sub.i >.sub.cell is established
for a combination of subbands of the conduction band and the
valence band that provides <f.sub.f .vertline.f.sub.i
>.apprxeq.1.notident.0. <u.sub.f .vertline.H.sub.I
.vertline.u.sub.i >.sub.cell is proportional to the extent of
polarization produced by excitation within the unit cell and
therefore to the lattice constant.
On the other hand, <f.sub.f .vertline.f.sub.i >=0 is true for
inter-subband transition because the envelope functions are
orthogonal each other, whereas <u.sub.f .vertline.u.sub.i
>.sub.cell .apprxeq.1 within the band because any Bloch
functions are substantially identical there. Thus, <u.sub.f
f.sub.f .vertline.H.sub.I .vertline.u.sub.i f.sub.i
>.apprxeq.<f.sub.f .vertline.H.sub.I .vertline.f.sub.i >is
obtained for any light in the TM mode that provides <f.sub.f
.vertline.H.sub.I .vertline.f.sub.i >.varies.=0. <f.sub.f
.vertline.H.sub.I .vertline.f.sub.i > is proportional to the
extent of polarization within the well and hence to the width of
the well. (See, inter alia, L. C. West and S. J. Englash, Appl.
Phys. Lett., vol.46, p.1156, 1985.)
The total thickness of the well layer 1031 and the intermediate
layers 1033a, 1033b is about 2.1 nm, which is more than three times
greater than the lattice constant of any conventional InGaAsP type
semiconductor. Since the nonlinear susceptibility factor
.chi..sup.(3) is proportional to the fourth power of the transition
matrix element, .chi..sup.(3) of the inter-band transition of this
embodiment is greater than that of the inter-band transition of any
conventional InGaAsP type substance by about a magnitude of two
digits. On the other hand, since the absorption coefficient .alpha.
is proportional to the square of the dipole moment, that of the
present invention is increased by a magnitude of one digit as
compared with the case of inter-band transition. This increase of
absorption can be compensated by the gain of the InGaAsP active
layer 1015 of the integrally formed waveguide.
As described earlier by referring to the second embodiment, while
the response time of inter-band transition is restricted by the
carrier lifetime, which is somewhere around 1 ns, the response time
of inter-subband transition is restricted by intra-band relaxation
time, which has a magnitude of 100 fs. Moreover, the time of
intra-band relaxation of a substance having a remarkable ionicity
and a large effective mass of electron such as GaN is expected to
be much shorter than that of an InGaAsP type substance. Thus the
response time of the embodiment is improved by a magnitude of three
to four digits relative to a conventional optical switch that
utilizes inter-band transition of an InGaAsP type substance.
As a result, the figure of merit, or .vertline..chi..sup.(3)
.vertline./(.alpha..tau.), of the nonlinear optical switch of the
above described third embodiment that utilizes inter-subband
transition is improved from that of a conventional nonlinear
optical switch that utilizes inter-band transition by a magnitude
of four to six digits and thus the embodiment can function as a
practical high speed high efficiency nonlinear optical switch that
operates at room temperature. Note that, while the optical switch
of the third embodiment is designed to operate with a wavelength of
1.55 .mu.m, an optical switch according to the present invention
and designed to operate with a longer wavelength may show a further
improvement in terms of greater nonlinearity because a thicker well
layer is used there.
By comparing the above embodiment with a conventional optical
switch comprising a transparent active waveguide, it will be seen
that the improvement in .alpha..tau. is small but the value of
.vertline..chi..sup.(3) .vertline. can be increased by a magnitude
of two digits so that the optical input energy required for
switching operation can be significantly reduced. Additionally, the
number of carriers within the well is maintained to a constant
level because the component semiconductor layers of the nonlinear
waveguide layer 1012 have a wide band gap and therefore it does not
give rise to multiple photon absorption relative to light having a
wavelength of 1.55 .mu.m nor induce injection of carriers from
other layers. Still additionally, since the optical input can be
held to a low level, it does not significantly affect the optical
nonlinearity of the layers other than the nonlinear waveguide layer
1012 nor does it give rise to any fluctuations in the number of
carriers due to two-photon absorption. Thus, consequently,
fluctuations in the gain that take place slowly for several
nanoseconds as well as fluctuations in the optical nonlinearity can
be effectively suppressed. As a combined effect of the advantages
pointed out above, the embodiment can operate stably even if a
quick pulse is applied repetitively.
The above described third embodiment can also be modified in a
number of ways as the resonant wavelength of intra-band transition
and the gain wavelength of the active waveguide layer resonate.
Additionally, the lattice constant and the crystal type may be
combined in a number of different ways through the use of direct
bonding. For instance, a wide band gap semiconductor substance such
as an SiCGe type material, II-VI group semiconductor or a
chalcopyrite material may be used in place of an InGaAlN type
substance for the inter-subband transition layer. If the influence
of two-photon absorption is not a problem, a quantum well of narrow
band gap semiconductor substance such as an AlGaInAsSb type
material may be used for the inter-subband transition layer. As in
the case of the second embodiment, inter-valence band transition
between the HeCdTe type spin orbit split-off band and the heavy and
light hole bands may be utilized. As described earlier by referring
to the second embodiment, the response time of inter-valence band
transition is very short because it is restricted by the intra-band
relaxation time. Still additionally, any appropriate semiconductor
material may be used for the active waveguide. Finally, an optical
coupler and a polarization coupler may be formed on Si substrate
and bonded to a nonlinear waveguide by direct bonding to produce an
integrated device. The type of substrate, the wavelength, the
manufacturing method, the electric current injection arrangement,
the optical waveguide structure and the well structure are not
limited to those described above for the third embodiment.
(4th Embodiment)
FIG. 18 is a schematic illustration of an optical control type
optical switch according to the fourth embodiment of the present
invention. The optical control type optical switch of this
embodiment is monolithically formed on an n-type InP substrate
101.
The optical switch comprises a first optical coupler 111, a second
optical coupler 112, a third optical coupler 113, a fourth optical
coupler, a fifth optical coupler that operates as signal light
branching means and a sixth optical coupler that operates as a
signal light sending out coupler, all of which are 1:1 directional
couplers (3 dB couplers).
Input optical signal is introduced into either one of the input
ports of the fifth optical coupler 115 via an input waveguide 135.
The fifth optical coupler 115 branches the input signal light to a
first intermediate optical waveguide (first intermediate light
path) 131 and a second intermediate optical waveguide (second
intermediate light path) 132 to a ratio of 1:1.
The first intermediate optical waveguide 131 and a first control
light input optical waveguide 141 are coupled to a first optical
waveguide 121 and a second optical waveguide 122, which is
structurally symmetrical relative to the first optical waveguide
121. The first and second optical waveguides 121 and 122 intersects
each other at a middle point and are then coupled to a first
control light output waveguide (first control light output light
path) 143 and a third intermediate optical waveguide (third
intermediate light path) 133 by the second optical coupler 112. The
stretch from the first optical coupler 111 to the second optical
coupler 112 constitutes a first Mach-Zehnder interferometer
151.
Likewise, the second intermediate optical waveguide 132 and the
second control light input optical waveguide 142 are coupled to a
third optical waveguide 123, which is structurally identical with
the first optical waveguide 121, and a fourth optical waveguide
124, which is structurally identical with the second optical
waveguide 122. The third and fourth optical waveguides 123 and 124
are coupled to a second control light output waveguide (second
control light output light path) 144 and a fourth intermediate
optical waveguide (reference light path) 134. The stretch from the
third optical coupler 113 to the fourth optical coupler 114
constitutes a second Mach-Zehnder interferometer 152, which has a
configuration identical with that of the first Mach-Zehnder
interferometer.
The third intermediate optical waveguide 133 and the fourth
intermediate optical waveguide 134 are coupled to a first output
optical waveguide (first signal light output light path) 136 and a
second output optical waveguide (second signal light output light
path) 137 by the sixth optical coupler 116. The stretch from the
fifth optical coupler 115 to the sixth optical coupler 116
constitutes a third Mach-Zehnder interferometer, which includes the
first Mach-Zehnder interferometer 151 and the second Mach-Zehnder
interferometer 152 disposed at the respective branching points.
The first through fourth optical waveguides 121, 122, 123, 124
respectively comprise active waveguide sections 161, 162, 163, 164,
each having a total length of 10 mm, and phase modulation sections
171, 172, 173, 174, each having a length of 500 .mu.m.
FIG. 19A is a schematic cross sectional view of the above
embodiment taking along the optical path from the input waveguide
135 all the way to the first output waveguide 136 via the fifth
optical coupler 115, the first intermediate optical waveguide 131,
the first optical coupler 111, the first optical waveguide 121, the
second optical coupler 112, the third intermediate optical
waveguide 133 and the sixth optical coupler 116. Note that the
optical coupling with the adjacent channel in the optical couplers
is neglected in FIG. 19A. Also note that a portion of the first
optical waveguide 121 forms an active waveguide section 161, while
another portion of the optical waveguide 121 forms a phase
modulator 171. Because of the structural symmetry, signal light
passing through the other branching points is made to pass optical
waveguides having an identical cross sectional view. The input and
output facets of each of the optical waveguides 135, 136, 137, 141,
142, 143 and 144 are coated with anti-reflection film 139 and
connected to the outside by optical fiber.
The waveguide layer of the embodiment basically comprises an n-type
InP substrate 101 that also operates as a clad layer, a common
passive waveguide layer 102 made of undoped InGaAsP and designed
for a PL wavelength of 1.2 .mu.m, an active waveguide layer 103
formed only in the active waveguide section 161 of the first
waveguide 121 and designed for a PL wavelength of 1.55 .mu.m, a
p-type InP clad layer 104 and a p-type InGaAsP ohmic contact layer
105. A common electrode 106 is arranged under the substrate.
Electrodes 215, 211, 212, 216 are respectively formed on the
contact layers of the optical couplers 115, 111, 112, 116 for
finely regulating the branching ratio to 1:1 by applying a reverse
bias voltage.
An electric current injection electrode 261 is formed on the
contact layer of the active waveguide section 161 to bias the
active waveguide in order to produce a transparent state. A reverse
bias electrode 271 is formed on the contact layer of the phase
modulation section 171 to regulate the phase of the Mach-Zehnder
interferometers. All the contact layer 105 and most of the p-type
InP clad layer 104 are removed from the optical waveguide except
where the electrodes are formed and then a semi-insulated InP layer
107 is formed there to fill the vacancy to electrically isolate the
upper electrodes.
FIG. 19B is a schematic cross sectional view of the above
embodiment taken along a plane perpendicular to the direction of
waveguide of the active waveguide sections 161, 162, 163, 164. Each
of the optical waveguides shows a mesa 108 having a width of 2
.mu.m and the gap between two adjacent mesas is filed with
polyimide 109 to produce a flat surface. Although not shown, each
of the upper electrodes is connected to a pad, which is then
connected to an external circuit by a bonding wire, whereas the
lower electrode is rigidly secured to a Cu block by AuSn solder,
said Cu block operating also as a heat sink and a ground.
The above optical control type optical switch operates in a manner
as described below. Assume here that signal light is a pulse having
a data rate of 100 Gb/s and a pulse width of 1 ps and optically
demultiplexed by a trapezoidal pulse of control light having a
frequency of 25 GHz and a pulse width of 5 ps. The peak power of
control light is so regulated that it shifts the phase of signal
light by .pi. in the first and second optical waveguides 121, 122.
As shown in FIG. 20, the two pulses are so regulated that the
plateau of each trapezoid of control light covers a peak of signal
light. The optical switch can respond to a pulse having such a high
frequency because it comprises active transparent waveguides.
Assume here also that the first through fourth active optical
waveguide sections 161, 162, 163, 164 are biased to provide
transparency with regard to the wavelength of optical input and
that the optical couplers 111, 112, 113, 114, 115, 116 are so
regulated by a reverse bias voltage applied thereto as to provide a
branching ratio of 1:1 and the Mach-Zehnder interferometers 151,
152, 153 are compensated for phase by the phase modulators 171,
172, 173, 174 to establish perfectly symmetry. Note that, the phase
of light crossing channels at any of the 1:1 optical couplers 111,
112, 113, 114, 115, 116 is shifted by .pi./2 relative to that of
light traveling through the channel.
If there is no control light, the signal light pulse is introduced
into the input waveguide 135 from an optical fiber and branched by
the fifth optical coupler 115 to the first intermediate waveguide
131 leading to the first Mach-Zehnder interferometer 151 and the
second intermediate waveguide 132 leading to the second
Mach-Zehnder interferometer 152 to a ratio of 1:1. The component of
light branched to the first intermediate waveguide 131 is further
branched by the first optical coupler 111 to the first optical
waveguide 121 and the second optical waveguide 122 to a ratio of
1:1. Since there is no control light and the signal light is weak,
no phase shift is produced to the signal light in the active
waveguide sections 161, 162 by the nonlinear optical effect. Thus,
the phase difference between the light coming from the first
optical waveguide 121 and the one coming from the second optical
waveguide 122 at the input section of the second optical coupler
112 is equal to .pi./2.
As a result of it, all the signal light output of the first
Mach-Zehnder interferometer 151 is coupled to the third
intermediate waveguide 133 to make the output to the first control
light output waveguide 143 equal to nil. Similarly, all the signal
light output of the second Mach-Zehnder interferometer 152 is
coupled to the fourth intermediate waveguide 134 by the fourth
optical coupler 114. The phase difference between the two
components of signal light introduced into the sixth optical
coupler 116 is still .pi./2, which is defined by the fifth optical
coupler 115. So, consequently, all the components of signal light
is given to the first output optical waveguide 136.
Assume now that control light is introduced into the first
Mach-Zehnder interferometer 151 from the first control light input
waveguide 141 by the first. optical coupler 111. Then, the control
light shifts the phase of signal light by .pi. in the first and
second optical waveguides 121, 122. Since the first Mach-Zehnder
interferometer 151 does not change the destination of signal light
for identical phase shifts in the two branches, all the components
of signal light there is introduced into the third intermediate
waveguide 133 although there is control light, while all the
control light is given to the first control light output waveguide
143. However, only the phase of the component of signal light
passing through the branch including the first Mach-Zehnder
interferometer 151 is shifted by .pi. in the third Mach-Zehnder
interferometer 153 and, therefore, the destination of signal light
is switched to the second signal light output waveguide 137.
As described above, the fourth embodiment can perfectly switch the
signal light output having a data rate as high as 100Gb/s from 0:1
to 1:0 and separate the control light and the signal light.
The present invention is not limited to the above embodiment. For
instance, if the first and second optical couplers are so arranged
that the length of either one is made equal to a 1/2 of the perfect
coupling length while that of the other one is made equal to 2/3 of
the perfect coupling length, the phase shift of the light crossing
the former will be .pi./2 and that of the light crossing the latter
will be -.pi./2 so that the crossing 125 will become unnecessary.
Alternatively, the crossing 125 may be eliminated by biasing either
the first optical waveguide 121 or the second optical waveguide 122
such that they produce a phase difference of .pi..
The above described functional features may be obtained by
combining fiber type optical couplers and optical fiber, although
the requirements of stability and downsizing may not necessarily be
met.
The present invention may find a number of different applications.
For example, control light may be reused with an optical switch
according to the present invention. If three optical switches 180a,
180b, 180c of the first embodiment are arranged for cascade
connection as shown in FIG. 21 so that control light and signal
light are introduced to each optical switch with a timing shifted
sequentially by 10 picoseconds by means of delay optical waveguide
181a, 181b, an optical demultiplexing from 100 Gb/s to 25 Gb/sx4
can be achieved. All the components may be monolithically and
integrally formed with a control pulse light source 182 and signal
receiving high speed waveguide type photodiodes 183a, 183b, 183c,
183d.
If an optical coupler is used for input signal branching means, a
two-input arrangement may be realized for signal light. If such is
the case, the outputs of two different signal lights may be
switched between cross and bar states. With such an arrangement, an
optically controlled ultrahigh speed optical switching operation
can be realized and applied to a self-routing switch for an optical
ATM switch.
In an optical switch having a second Mach-Zehnder interferometer
may be used for optical logic operations by using a first control
light introduced from a first control light input waveguide and a
second control light introduced from a second control light input
waveguide. If the phase of the output signal light of the second
Mach-Zehnder interferometer is shifted by .phi. by the second
control light, the output optical coupler of the third Mach-Zehnder
interferometer will produce a phase difference of .phi.--.phi..
Thus, signal outputs corresponding to exclusive OR and its negation
for two control lights can be obtained at the output of the third
Mach-Zehnder interferometer by utilizing this phenomenon. The
second control light may be used for phase biasing by using
continuous light for it.
While the present invention is described above exclusively in terms
of digital routing switching operations, the ratio of two outputs
may be modified to any given ratio by continuously regulating the
input control light. Therefore, the present invention may be
applied to an ultrahigh speed optical control type optical
modulator. FIG. 22 is a graph showing the relationship between the
input optical power and the output when the optical switch of the
fourth embodiment is used for an analog optical modulator. It will
be seen that complicated modulating operations can be carried out
for signal light by using two input control lights.
It will also be seen that the present invention has a number of
different applications.
As described above in detail by referring to the first through
fourth embodiments, the nonlinearity attributable to intra-band
absorption can be made greater than the nonlinearity attributable
to two-photon absorption by using a material that makes the
resonant wavelength of intra-band absorption substantially equal to
the wavelength of incident light for at least part of the layered
structure of optical waveguide. Consequently, a device according to
the present invention can realize a high frequency switching
operation with a low consumption rate of excitation energy more
efficiently than any comparable conventional devices. Additionally,
a device according to the present invention can perfectly switch
the destination of output signal light with a large extinction
ratio so that a high speed high efficiency optical control type
optical switch can be realized.
For reference, expected advantages of the optical switch according
to the present invention is shown in terms of various parameters in
FIG. 33.
(5th Embodiment)
FIG. 23 is a partially cut-away schematic perspective view of a
wavelength conversion device according to the fifth embodiment of
the present invention.
In the wavelength conversion device, four wave mixing of a
traveling wave type semiconductor laser amplifier is used for
wavelength conversion. Referring to FIG. 23, a stripe-shaped
semiconductor optical waveguide 302 is formed on an n-type InP
substrate 301 that also operates as a n-type clad layer. The
semiconductor optical waveguide 302 is realized by sequentially
arranging an n-type clad layer (n-type InP substrate 301), an
undoped InGaAsP optical waveguide layer 303, a strained quantum
well active layer 304 comprising an undoped strained
InGaAs/strained AlAs quantum well, an InGaAsP optical waveguide
layer 305 and a p-type InP clad layer 306 in the above order to
form a multilayer structure. A p-type InGaAsP ohmic contact layer
307 is formed on the p-type InP clad layer 306.
A buried layer comprising a p-type InP layer 308 and an n-type InP
layer 309 is formed along the lateral sides of the semiconductor
optical waveguide 302 in order to confine an electric current to
the semiconductor optical waveguide 302. A p-side ohmic electrode
311 and an n-side ohmic electrode 312 are arranged respectively on
the p-type InGaAsP ohmic contact layer 307 and under the InP
substrate 301. The input and output facets are coated with an
anti-reflection film 313 to suppress the reflectivity of the facets
to less than 0.1%.
The wavelength conversion device is arranged on a Cu mount (not
shown) provided with an Au coated AlN submount that also operates
as a heat sink and connected to feed lines by bonding.
The wavelength conversion device is typically assembled with
input/output optical fibers, a pair of aspherical lenses for
realizing a low loss optical coupling with the input/output optical
fibers, an optical isolator and a Peltier cooler to produce a
module. As shown in FIG. 6, this module may be combined with an
optical multiplexer/demultiplexer, a narrow band optical filter and
an optical amplifier to carry out wavelength conversion of light
that operates as a signal carrier wave.
In the fifth embodiment, the strained quantum well active layer
operates as a semiconductor layer having a resonant wavelength of
intra-band absorption defined within the gain band of a traveling
wave type semiconductor laser amplifier. FIG. 24 shows a schematic
illustration of the structure of the conduction band of a principal
portion of the strained quantum well active layer 304 of the
semiconductor optical waveguide 302.
The strained quantum well active layer 304 is realized by arranging
twenty (20) unit layers, each comprising a thin InGaAs well layer
314 and a thin tensile strained AlAs barrier layer 315, the InGaAs
well layer being sandwiched between a pair of barrier layers. There
exists a first subband 316 and a second subband 317 in the InGaAs
well layer 314.
The transition from the first subband 316 to the second subband 317
is an allowed transition based on the parity rule. Since the
tensile strained AlAs barrier layer 315 is thin, the subbands 316
and 317 of each InGaAs well layer 314 are coupled by tunneling to
form minibands. The inter-subband transition energy is about 0.8 eV
(a resonant wavelength of 1.55 .mu.m).
That such a large gap can be formed between subbands is described
in J. H. Smet et al., Appl. Phys. Lett., vol.64, pp.986-987,
1994.
An electric current is injected into the strained quantum well
active layer 304 in such a way that it may give rise to a
sufficient gain to light having a wavelength of or close to 1.55
.mu.m. Thus, electrons are injected at a high density into the
first subband 316 by way of the tunneling, whereas holes are
injected into the valence band.
Because of the population inversion of electrons and holes, a
stimulated emission gain is produced in the strained quantum well
active layer 304 so that light having a wavelength of or close to
1.55 .pi. m and introduced into the semiconductor optical waveguide
302 is amplified.
If pump light having an angular frequency of .omega..sub.1 and
signal light having an angular frequency of .omega..sub.2
=.omega..sub.1 -.OMEGA. are introduced simultaneously, the gain and
the refractive index are modulated by a beat frequency of .OMEGA.
and a light with an angular frequency of .omega..sub.3
=.omega..sub.1 +.OMEGA. is generated as in the case of a
conventional traveling wave type semiconductor laser amplifier so
that only the component with the angular frequency .omega..sub.3
can be taken out by means of an external narrow band optical filter
as in the case of a conventional wavelength conversion device.
Since the absorption resonant wavelength between the first and
second subbands 316 and 317 of the conduction band locates in the
gain wavelength, electrons in the first subband absorb part of the
light in the semiconductor optical waveguide and excited to the
second subband 317, as shown in FIG. 25.
The electrons excited to the second subband relax to the high
energy level of the first subband in a short period of time as they
collide with phonons. Then, the electrons at the high energy level
relax to the original low energy level as they collide with
electrons and phonons (intra-band carrier relaxation).
The process of intra-band carrier relaxation is basically identical
with the relaxation process of spectral hole burning and that of
carrier heating and has a time constant as short as hundreds of
several femtoseconds to several picoseconds.
It will be appreciated that such a remarkable extent of
nonlinearity cannot be achieved by any conventional wavelength
conversion device, where only nonlinear effects such that free
carrier absorption and two-photon absorption modulate the
intra-band carrier energy distribution except spectral hole burning
due to the stimulated emission.
With the intra-band resonant absorption of the above described
fifth embodiment, to the contrary, the refractive index and the
gain are modulated to a large extent, since the electron energy
distribution is remarkably changed by the optical field variation
with a beat frequency .OMEGA..
By adding the effect of four wave mixing of intra-band absorption
(complex coupling efficiency C.sub.4, time constant .tau..sub.4) to
formula (2), the conversion efficiency can be expressed by the
equation below. ##EQU2##
As described above, the relaxation process is a complex process
involving several different time constants and, therefore, needs to
be expressed by adding a fifth effect (C.sub.5, .tau..sub.5) and
further subordinate effects to the above formula to make it more
accurate.
However, if the active layer itself is the inter-subband resonant
absorption layer as in the case of the fifth embodiment, such
additional effects may be regarded not as new effects attributable
to inter-subband transition but as revised values of C.sub.2 and
C.sub.3 increased as a result of intra-band resonant
absorption.
For the purpose of simplification, the component having a time
constant raised to the level of that of carrier heating as a result
of intra-band resonant absorption is expressed by C.sub.4,
.tau..sub.4 hereinafter.
FIG. 26 is a graph showing the relative wavelength conversion
efficiency (solid line) of the wavelength conversion device of the
fifth embodiment in comparison with that (broken line) of a
conventional device.
It is seen from FIG. 26 that a high relative wavelength conversion
efficiency can be achieved with a device according to the present
invention even when the wavelength difference exceeds 1 nm. In
other words, a device according to the present invention can
realize highly efficient wavelength conversion over a bandwidth
broader than ever. This is because the present invention can
provide a large absolute value for the complex coupling efficiency
C.sub.4 and a value substantially equal to that of .tau..sub.2 for
the time constant .tau..sub.4. Therefore, a device according to the
present invention can respond to a signal that is modulated to a
high data rate of tens of several Gb/s and can be used for optical
demultiplexing of optical time-division multiplexed signals by
means of a short optical control pulse having a pulse width as
small as 1 ps.
Inter-subband transition is normally allowed for the TM mode and
forbidden for the TE mode. However, it is known that, if the
transition energy is large, absorption is also observable in the TE
mode because of possible deviation of the dispersion curve of the
conduction band from a parabolic curve, reduced symmetry due to
strain and other reasons.
Thus, various combinations may be possible for polarized waves of
exciting light, signal light and conjugate light so that desirable
values may be selected for the absorption coefficient and the
nonlinear susceptibility. A polarization coupler may conveniently
be used for light multiplexing by combining orthogonally polarized
light waves.
Since energy dispersion curves against the wave number in the well
plane for each subband of the conduction band are substantially
parallel to each other, the half width of an inter-subband
absorption spectrum is normally small.
However, with the quantum well of the fifth embodiment, the half
width is made rather wide as a result of the formation of minibands
and the dispersion curve is deviated from the parabolic curve so
that resonant absorption can be realized over a wide range of
wavelength in practical applications.
Note that, while the net gain may be reduced for the semiconductor
laser amplifier as a result of inter-subband absorption, the
influence of such reduction in the gain can be compensated by
connecting an external optical amplifier.
As described above, the wavelength conversion device of the fifth
embodiment can realize a high conversion efficiency. Thus, the
level difference between exciting light and conjugate light and the
level difference between signal light and conjugate light that are
very large in a conventional wavelength conversion device can be
reduced significantly, and the level difference between noise of
the semiconductor laser amplifier and conjugate light can be
increased. Therefore, the extinction ratio may not necessarily be
rigorously defined for a narrow band optical filter if the fifth
embodiment is used. Additionally, it can greatly improve the S/N
ratio.
(6th Embodiment)
FIG. 27 is a partially cut-away schematic perspective view of a
wavelength conversion device according to the sixth embodiment of
the present invention.
In this embodiment, four wave mixing of a traveling wave type
semiconductor laser amplifier is utilized for wavelength
conversion.
The sixth embodiment of semiconductor optical waveguide device is a
traveling wave type semiconductor wavelength amplifier comprising
an InGaAsP active layer and a GaN/AlN quantum well layer (second
optical waveguide layer) that operates as part of an optical
waveguide. The intersubband absorption wavelength of the GaN/AlN
quantum well layer can be controlled by applying an electric
field.
Referring to FIG. 27 comprises a stripe-shaped optical waveguide
402 formed on a p-type InP substrate 401. The semiconductor optical
waveguide 402 is realized by sequentially arranging an InP
substrate 401 that also operates as a lower clad layer, a p.sup.-
-type InGaAsP optical waveguide layer 403, an undoped
InGaAs/InGaAsP quantum well active layer 404, an n-type InGaAsP
optical waveguide layer 405, an undoped InGaN layer 411, a quantum
well layer 412 operating as an inter-subband transition resonant
absorption layer and made of undoped GaN/n-type AlN and an n-type
AlGaN layer 413 that operates as an upper clad layer in the above
order to form a multilayer structure.
The optical waveguide 402 is externally surrounded by an n-type InP
layer 408 and a p-type InP layer 409 that operate as electric
current confining layers as well as by an n-type InGaAsP layer 407
arranged so as to contact the lateral sides of the active layer
404. The GaN/AlN quantum well layer 412 includes a region having a
relatively large well width and another region having a relatively
small well width, which regions are arranged along the waveguide. A
pair of grooves 414, 414 are formed along the respective lateral
sides of the optical waveguide 402 to produce a ridge-shaped
optical waveguide.
Electrodes 415, 416, 417 are formed respectively on the n-type
InGaAsP layer 407, on the n-type AlGaN layer 413 and under the
p-type InP layer 401. The input and output facets are coated with
an anti-reflection film 418 and the optical waveguide 402 is
separated with the facets by the window structure, so that the
reflectivity of the facets are suppressed to be less than 0.1%.
The above structure can be typically prepared in a manner as
described below. Firstly, an optical waveguide layer 403, an active
layer 404 and an optical waveguide 405 are sequentially formed by
epitaxial growth on a substrate to produce a semiconductor
epitaxial growth substrate 401. Apart from this, a nitride
epitaxial multilayer film is formed on a substrate which is
typically made of sapphire by arranging an InGaN layer 411, an
inter-subband absorption layer 412 and an AlGaN layer 413 via an
ZnO layer or a buffer layer disposed therebetween. A well width
modulation structure can be produced by repeating a selective
growth process twice for the respective regions, although such a
well width modulation structure may alternatively be prepared
through a single growth process by using a technique such as mask
selective growth (capable of changing the growth rate by mask
width) which is well known for the growth of InGaAsP.
Thereafter, the substrate and the nitride epitaxial multilayer film
are separated from each other by selectively etching the ZnO layer.
The obtained epitaxial multilayer film is then bonded to an InP
substrate 401 under pressure in such a way that the InGaAsP optical
waveguide layer 405 and the undoped InGaN layer 411 are arranged
vis-a-vis, and the assembled components are subjected to a heat
treatment in a hydrogen atmosphere to produce a unified entity. The
bonding operation normally shows an excellent result because both
of the oppositely arranged layers contain indium (In).
The directly bonded multilayer structure prepared in this way is
then dry-etched to produce a designed profile of optical waveguide
402 and, thereafter, buried layers 408, 409, 407 are formed by
epitaxial growth, using the nitride epitaxial multilayer film as a
selective growth mask. Subsequently, a pair of grooves 414, 414 are
formed on the respective lateral sides of the waveguide. Then,
upper electrodes 415, 416 are prepared and the underside of the InP
substrate 401 is polished to form a lower electrode 417 thereon.
Thereafter, a device as illustrated in FIG. 27 is produced as a
result of a series of processing operations including cleavage and
dicing for cutting out chips and formation of anti-reflection films
418.
A semiconductor chip prepared in this manner is then put on a Cu
mount having an Au coated AlN submount that operates as a heat sink
and an electric terminal with the lower electrode 417 disposed
therebetween, whereas the upper electrodes are connected to feed
lines by bonding and by way of strip lines. Then, it is put
together with input/output optical fibers, a pair of aspherical
lenses for realizing a low loss optical coupling with the
input/output optical fibers, an optical isolator and a Peltier
cooler to produce a module.
With a wavelength conversion device as illustrated in FIG. 27,
electrons are injected from the upper electrode 415 into the active
layer 404 by way of the InP buried layer 407 and the InGaAsP layer
405, whereas holes are injected from the substrate 401 into the
active layer 404 through the InGaAsP layer 403. Since a stimulated
emission gain is produced in the active layer 404 as a result of
population inversion of electrons and holes, light having a
wavelength of or close to 1.55 .mu.m and introduced into the
optical waveguide 402 is amplified. As no multiple reflection takes
place along the direction of optical waveguide, no laser
oscillation occurs during the operation of current injection and
amplification is realized with a large gain.
Since the bandgaps of InGaN, GaN and AlN are by far greater than
those of InP and InGaAsP and there is a high potential barrier for
the carriers in the active layer 404 and the optical waveguide
layer 405, no carrier injection takes place there. Additionally,
since the bandgap of GaN is large (3.4 to 3.6 eV), none of two-,
three- and four-photon absorption takes place as a result of
inter-band transition if light with a wavelength of 1.55 .mu.m band
is introduced so that the carrier density in the GaN well layer
does not significantly fluctuate.
A deep quantum well is formed in the conduction band of the GaN/AlN
quantum well layer 412.
The bottom of the conduction band of AlN and that of the conduction
band of GaN are found at point .GAMMA..sub.1. While AlN that
constitutes a barrier layer for the quantum well layer 412 is doped
to n-type, most electrons are distributed in the first subband
related to .GAMMA..sub.1 of GaN when there is no light. An undoped
AlN layer may be arranged between the n-type AlN barrier layer and
the undoped GaN well layer in order to prevent the impurities of
AlN from adversely affecting the potential of the hetero interface.
If the most externally located AlN barrier layers are undoped and
have a sufficient thickness, they can prevent a real electric
current from flowing therethrough when a voltage is applied
thereto.
In regions of the GaN/AlN quantum well layer where the well has a
relatively large width, it is so regulated as to make the energy
difference between the first and second subbands of .GAMMA..sub.1
equal to 0.79 eV. On the other hand, in regions where the well has
a relatively small width, it is so regulated as to make the energy
difference between the first and second subbands equal to 0.81 eV.
The transition between the first and second subbands is allowed for
the TM mode.
Since energy dispersion curves against wave number in the well
plane of the two subbands are separated by a constant energy
regardless of the wave number at and near .GAMMA..sub.1 where
electrons exist, the width of the absorption spectrum can be
reduced to as small as 20 meV if possible causes of uneven energy
spread such as fluctuations in the well width and those in the
potential due to impurities can be effectively eliminated.
While light shows an energy distribution peak in or near the active
layer 404, it is in fact guided in a mode that allows it to partly
permeate into the GaN/AlN quantum well layer 412. Therefore, if
light having a wavelength corresponding to the above described
inter-subband transition (1.53-1.57 .mu.m) is transmitted through
the waveguide 402 in the TM mode, part of the light is absorbed by
electrons in the first subband to excite them to the energy level
of the second subband.
Four wave mixing takes place under this condition as the absorption
coefficient and the refractive index fluctuate depending on the
intensity of the guided light.
Thus, the wavelength conversion device of the sixth embodiment
operates basically same as the fifth embodiment except that
intra-band resonant absorption takes place in an inter-subband
absorption layer formed independently from the active layer.
Therefore, it can efficiently operate for wavelength conversion
over a large bandwidth.
(7th Embodiment)
A wavelength conversion device according to the seventh embodiment
has a configuration substantially same as that of the traveling
wave type semiconductor laser amplifier according to the first
embodiment (FIG. 8) described earlier. So, those component of the
seventh embodiment that are same or similar to their counterpart of
the first embodiment will not be described here any further. While
the first embodiment of semiconductor optical waveguide device is
designed as a traveling wave type semiconductor laser amplifier,
the seventh embodiment is used as a wavelength conversion device.
The phenomenon of four wave mixing of a traveling wave type
semiconductor laser amplifier is used for wavelength conversion in
the seventh embodiment.
Referring to FIG. 8, an electric current is injected into the
Hg.sub.0.3 Cd.sub.0.7 Te active layer 24 by way of the n-side and
p-side electrodes 26 and 27 of the seventh embodiment. The
Hg.sub.0.3 Cd.sub.0.7 Te active layer 24 shows a stimulated
emission gain at or near the wavelength of 1.3 .mu.m. Note that the
Hg.sub.0.3 Cd.sub.0.7 Te active layer 24 of this embodiment is a
semiconductor layer designed to have an inter-valence band
absorption wavelength found within the gain band of the traveling
wave type semiconductor laser amplifier.
Therefore, both amplification due to the stimulated emission gain
and inter-valence band absorption take place, and both the
refractive index and the gain change remarkably as the energy
distribution of holes changes to consequently give rise to a high
degree of nonlinearity and a high conversion efficiency.
Additionally, the time required for relaxation of inter-valence
band absorption is as short as hundreds femtoseconds at most. Thus,
the conversion efficiency does not drop remarkably if the
wavelength difference is increased.
As will be understood from the above description, the seventh
embodiment can be used for high efficiency wavelength conversion
over a large bandwidth as in the case of the fifth and sixth
embodiments that utilize inter-subband transition in the conduction
band.
Note that the present invention is by no means limited to the above
described embodiments. For instance, the materials and the
compositions of the active layer and the intra-band resonant
absorption layer and their thicknesses as well as the structure of
the semiconductor optical waveguide are not limited to those
described above by referring to the embodiments. In other words,
the gain wavelength, the absorption coefficient, the extent of
nonlinearity and the polarization can be modified by structurally
modifying the device. The semiconductor layer, in which the
resonant wavelength of intra-band absorption is found within the
gain band of the traveling wave type semiconductor laser amplifier,
operates as an active layer or it is independent from the active
layer and the clad layer in the above description. However, the
semiconductor layer may alternatively operates as part of the clad
layer. What is essential for the semiconductor layer is that it is
located within the power distribution zone of light guided through
the semiconductor optical waveguide.
Additionally, the wavelength conversion device is not necessarily
required to be an independent device. It may be combined with one
or more than one semiconductor lasers, optical modulators, optical
switches, optical multiplexer/demultiplexers, wavelength selection
devices, light receiving devices and other wavelength conversion
devices to form a larger integrated entity.
In short, as described above by referring to the fifth through
seventh embodiments, a large bandwidth high efficiency wavelength
conversion device can be realized by using a semiconductor optical
waveguide having a semiconductor layer whose intra-band absorption
resonant wavelength is located within the gain band of a
corresponding traveling wave type semiconductor laser
amplifier.
(8th Embodiment)
A semiconductor optical waveguide device according to the eighth
embodiment is a tunable wavelength filter having a configuration
substantially equal to that of the wavelength conversion device
according to the sixth embodiment (FIG. 27). So, those component of
the eighth embodiment that are same or similar to their counterpart
of the sixth embodiment will not be described here any further.
While the sixth embodiment of semiconductor optical waveguide
device is designed as a wavelength conversion device, the eighth
embodiment is used as a tunable wavelength filter.
FIG. 28A shows a graph of the spectrum of the gain (dotted broken
line) of the active layer of FIG. 27 and the that of the loss
(broken line) of the GaN/AlN quantum well layer 412 when no voltage
is applied thereto and FIG. 28B shows the corresponding net
transmission spectrums. The loss of the transmission band caused by
the inter-subband absorption is compensated by the gain of the
active layer 404.
As a voltage is applied between the electrodes 415 and 416 to apply
an electric field to the quantum well, the inter-subband energy
difference is enlarged by the quantum confining Stark effect to
shift the resonant wavelength to the shorter side. If the GaN/AlN
quantum well 412 has an asymmetric structure such as
AlN/GaN/AlGaN/AlN, large changes in the inter-subband resonant
absorption wavelength can be produced by applying an electric field
although the absorption spectrum comes to show a large width. The
broken line in FIG. 28B indicates the net gain/loss spectrum
obtained by applying a voltage. The rate of the transmission
wavelength change caused by an electric field is limited mainly by
LCR of the voltage applying system so that a response time of
several hundred picoseconds can be obtained.
Note that, while the two regions (a region having a relatively
large well width and a region having a relatively small well width)
of the second semiconductor optical waveguide layer of the eighth
embodiment are simultaneously controlled by a common electrode 416,
strictly speaking, they are different from each other in terms of
the extent of shift of the absorption peak wavelength relative to
the applied voltage. Additionally, the width of the absorption
spectrum and the absorption coefficient also change as a function
of the voltage. In order to effectively control the transmission
peak wavelength regardless of these changes, keeping the
transmission characteristics substantially constant, it is
preferable to divide the electrode 416 into two sections so that
the above two regions may be controlled independently. With an
electrode having two divided section, it is also possible to
control certain aspects of transmission other than the transmission
wavelength such as the transmission bandwidth.
A filter having a plurality of transmission wavelengths or a
tunable wavelength filter having a complex distribution pattern of
transmission and absorption bands can be realized by dividing the
second semiconductor waveguide layer 412 into a number of regions
greater than two.
The tunable wavelength filter of the eighth embodiment is protected
against multiple reflections of light by the anti-reflection film
418 formed on the facets and the window structure so that the width
of the short optical pulse may not be expanded. Electrons excited
to the energy level of the second subband quickly relax to the
original low energy level in a very short period of time of several
picoseconds as they repeatedly collide with electrons and phonons.
Therefore, the foregoing pulse would not affect the following pulse
even if the energy of short optical pulses is absorbed for every
several picoseconds.
As described below, the tunable wavelength filter of the eighth
embodiment differs from a device in which a light absorbing section
for inter-subband absorption that can quickly tune the wavelength
and an optical amplifier section that is free from inter-subband
absorption are connected with a cascade connection arrangement.
With an arrangement where an optical amplifier is connected behind
a light absorbing section, ASE noises generated by the optical
amplifier is added to the absorption wavelength to worsen the
signal to noise ratio. The signal with an aggravated S/N ratio
cannot restore its original signal quality by amplification.
Contrary to this, the signal to noise ratio of a semiconductor
optical waveguide device according to the invention can be
maintained to a high level because ASE noises in the absorption
wavelength band are also absorbed. A semiconductor optical
waveguide device according to the present invention has an
amplifying ability and, if used as an initial amplifier having an
excellent signal to noise ratio and connected to a downstream
optical amplifier, higher output power can be obtained.
To the contrary, in an arrangement where a light absorbing section
utilizing inter-subband absorption is disposed downstream and
connected to an optical amplifier, light is firstly amplified by
the optical amplifier and then introduced into the light absorbing
section. This arrangement cannot provide outputs stably since it is
affected by saturation of inter-subband absorption, incidental heat
generation and enhanced nonlinearity. It will be seen from the
above that a semiconductor optical waveguide device according to
the present invention can provide a wide dynamic range for the
input and output energy levels.
FIG. 29 is a schematic illustration of a wavelength conversion node
realized by using a tunable wavelength filters of FIG. 27. The
major components of the wavelength conversion node include a
rapidly wavelength tunable semiconductor laser 441, a coupler for
multiplexing signal light and the optical output of the tunable
semiconductor laser 441, a wavelength conversion device 443
consisting in a traveling wave type semiconductor laser amplifier,
a semiconductor optical waveguide device (tunable wavelength
filter) 444 according to the present invention, an optical fiber
amplifier 445 and a control unit 446.
For the wavelength conversion node 440 of FIG. 29, signal light is
constituted by a packet of short pulse train having a wavelength of
.lambda..sub.q. The rapidly wavelength tunable semiconductor laser
442 produces a pump light pulse having a wavelength of
.lambda..sub.p defined for each packet in synchronism with the
signal light pulse. The signal light having the wavelength of
.lambda..sub.q and the pump light having the wavelength of
.lambda..sub.p are subjected to four wave mixing in the wavelength
conversion device 443, which generates conjugate signal light
having a wavelength of .lambda..sub.c =2.lambda..sub.p
-.lambda..sub.q.
The control unit 446 operates to control the oscillation wavelength
of the rapidly wavelength tunable semiconductor laser 441 to make
it agree with the wavelength .lambda..sub.p determined by the
wavelength .lambda..sub.q of the original signal packet and the
wavelength .lambda..sub.c of the packet to be obtained by
conversion. It also controls the transmission wavelength
.lambda..sub.c of the rapidly wavelength tunable filter 444. Thus,
the operation of wavelength conversion can be carried out for each
packet.
Routing of the signal light can be realized by connecting a planer
lightwave circuit (e.g., planer lightwave circuit (PLC)) that
defines the output as a function of wavelength to this node. This
node may also be applicable to an add-drop multiplexer. A space and
wavelength division multiplexed optical switch can be realized by
arranging a plurality of such nodes in parallel and connecting them
via routing PLC to produce a multi-stage arrangement.
Since conventional wavelength conversion nodes are slow in
response, they cannot be used for a high speed packet switch while
they are feasible for the operation of slow wavelength switching
type cross connect that entails a long switching time. Thus, a
semiconductor optical waveguide device according to the present
invention will be able to develop a number of new applications
including those described above.
While an optical waveguide device according to the present
invention is used for the rapidly wavelength tunable filter 444 of
FIG. 29, it may also be used for the wavelength conversion device
443, which is a traveling wave type semiconductor laser
amplifier.
More specifically, if the inter-subband absorption wavelength is
close to any of the wavelengths involved in four wave mixing, the
efficiency of four wave mixing caused by carrier heating or
spectral hole burning can be improved further. Additionally, the
wavelength conversion performance can be optimized for each
combination of wavelengths involved in the operation of wavelength
conversion by externally controlling the inter-subband absorption
spectrum.
For example, there can be a combination of wavelengths with which
the efficiency of wavelength conversion can be reduced as a result
of phase interference of the three factors of carrier density
change, carrier heating and spectral hole burning that participate
in the operation of four wave mixing. If such is the case, the
conversion efficiency can be improved as a result of interference
by changing the ratio (a parameter) of the real part and the
imaginary part of the nonlinear susceptibility .chi..sup.(3) by
applying a voltage to the second semiconductor optical waveguide
layer so that the operation of wavelength conversion can be carried
out highly efficiently regardless of the combination of
wavelengths.
In short, a semiconductor optical waveguide device according to the
present invention is a multifunctional device that can be used as a
wavelength conversion device or a tunable wavelength filter. In
other words, an integrated entity comprising a number of wavelength
conversion devices and tunable wavelength filters may be produced
by a single manufacturing process. This sort of integration
provides a number of advantages including the following. (1) The
coupling loss involved in connecting the devices with optical
fibers can be reduced. (2) An enhanced stability relative to
changes in the environment can be achieved. (3) No additional
effort for modularization is required. (4) A small and lightweight
unit can be manufactured with ease. (5) A significant cost
reduction can be realized. Additionally, most part of the
wavelength conversion node of FIG. 29 can be produced in the form
of a single chip because a semiconductor optical waveguide device
according to the present invention can be used for a tunable
wavelength laser as will be described hereinafter by referring to
the ninth embodiment of the present invention.
(9th Embodiment)
A semiconductor optical waveguide according to the ninth embodiment
of the present invention is applied to a tunable wavelength DFB
laser. FIG. 30 is a schematic cross sectional view of a tunable
wavelength DFB laser, shown along the optical waveguide.
Referring to FIG. 30, the tunable wavelength DFB laser comprises a
p.sup.- -type InGaAsP optical waveguide layer 453, an undoped
tensile strained InGaAsP/InGaAsP quantum well active layer 454, an
n-type InGaAsP optical waveguide layer 455, an undoped InGaN layer
461, an undoped GaN/n-type AlN quantum well layer 462 for operating
as an inter-subband absorption layer and an n-type AlGaN layer 463
for operating as an upper clad layer are arranged sequentially on a
p-type InP substrate 451 to form a multilayer structure. The basic
configuration of the above structure is same as that of the tunable
wavelength filter of the above described eighth embodiment. The
embodiment shows a cross sectional view similar to that of the
tunable wavelength filter of the eighth embodiment along a plane
perpendicular to the optical waveguide 452 having the above
multilayer structure.
Electrodes 465 and 467 are formed respectively on the n-type AlGaN
layer 463 and under the substrate 451, while the n-type InGaAsP
optical waveguide 455 is electrically connected to a third
electrode (not shown) as in the case of the tunable wavelength
filter of the eighth embodiment. The procedures for injecting an
electric current and for applying a voltage are also similar to
those of the tunable wavelength filter of the eighth embodiment. An
optical isolator is incorporated into the module to prevent any
externally reflected light from coming back to the laser.
The tunable wavelength DFB laser of FIG. 30 differs from the
tunable wavelength filter of the eighth embodiment in that a
diffraction grating is formed on the interface of the InP substrate
451 and the InGaAsP optical waveguide layer 453, that the active
layer has a strained quantum well structure and that the active
layer does not have a plurality of regions along the optical
waveguide. A 1/4 wavelength phase shifter 471 is formed at the
center of the diffraction grating 470 whereas an anti-reflection
film 472 is formed on each of the facets. The number of wells in
the GaN/AlN layer 462 is slightly larger than that of the eighth
embodiment so that the optical mode distribution extensively
overlies the well layer.
Since optical feedback is realized by the diffraction grating, a
single mode oscillation can occur when an electric current is
injected into the active layer 454 at sufficiently high rate.
Thanks to the phase shifter 471 and the anti-reflection films 472,
the oscillation proceeds stably at and near the center of the stop
band around the Bragg wavelength. Since the active layer 454
comprises tensile strained quantum wells, the oscillation takes
place in the TM mode.
As shown in the graph of FIG. 31, the refractive index for the TM
mode is high at the long wavelength side of the wavelength band
that give rise to inter-subband absorption and low at the short
wavelength side. While the absorption spectrum has a narrow width,
the refractive index fluctuates over a wide range of wavelength as
a result of inter-subband absorption. The rate at which the
refractive index change is reduced as the extent of detuning with
the inter-subband absorption wavelength increases.
Thus, by appropriately selecting the extent of detuning between the
inter-subband absorption wavelength without electric field and the
laser oscillation wavelength, the refractive index can be reduced
by electric field without remarkably modifying the absorption
coefficient. Additionally, since optical mode distribution
extensively overlies the GaN well layer, the reduction in
equivalent refractive index of the optical waveguide is large and
the oscillation wavelength is greatly shifted to the short
wavelength side.
Since the response speed for a change in the wavelength is
subjected to LCR restrictions, a wavelength tuning operation can be
carried out with ease in a sub-nanosecond range if sufficient care
is taken for the drive circuit and the mounting method. Note here
that it is very difficult to change only the refractive index with
inter-band transition because the absorption spectrum is normally
wide. To the contrary, fluctuations in the output power can be
minimized according to the present invention because the refractive
index can be varied without greatly modifying the absorption
coefficient.
While the present invention is described above in terms of
application to a tunable wavelength laser, it can also be applied
to a Fabry-Perot type semiconductor laser having reflection planes
at the opposite facets to realize mode locking by utilizing the
second semiconductor optical waveguide layer as a saturable
absorber.
Also, as shown in FIG. 32A, the second semiconductor optical
waveguide layer 491 may be formed only in part of the waveguide
492. Alternatively, as shown in FIG. 32B, the region containing the
active layer 493 and the region containing the second semiconductor
optical waveguide layer 491 may be connected in series.
Additionally, a sine wave voltage may be applied to the voltage
terminal 494 to control absorption in order to realize active mode
locking. By so arranging that the resonant wavelength can be
modified on the both sides of the wavelength of the center of
oscillation by applying a sine wave voltage, the active mode
locking can be realized with a period twice as large as that of the
modulation voltage. The performance of the laser can be controlled
by means of a cw-voltage in the case of passive mode locking.
If the second semiconductor optical waveguide layer is integrally
formed outside the resonator of the DFB laser, the device can be
used for an optical modulator integration type light source. Then,
as shown in FIG. 31, the ratio (.alpha. parameter) of the change in
the refractive index and the change in the absorption coefficient
can be modified to a great extent by selecting appropriate values
for the wavelength and the bias.
The present invention is not limited to the above embodiments and
various modifications can be made to them. More specifically, while
GaN/AlN quantum wells are used for the second semiconductor optical
waveguide layer in the above description, inter-subband absorption
can be realized for a wavelength of 1.55 .mu.m in an
InGaAs/strained AlAs conduction band or a II-VI group semiconductor
valence band. When InGaAs/strained AlAs is concerned, both the
first and second semiconductor optical waveguide layers may be
produced in a single epitaxial growth process. Note that the
wavelength is not necessarily limited to 1.55 .mu.m.
Inter-subband absorption of valence band or intervalence band
absorption can be utilized for intra-band resonant absorption. In
order to avoid two-photon absorption from taking place, it is
preferable to use a wide band gap semiconductor such as GaN,
although, conversely, it may be possible to realize new nonlinear
optical devices by combining the effect of two-photon absorption in
the second semiconductor optical waveguide and that of inter-band
resonant absorption. A variety of combinations of polarized input
and output lights may be possible for nonlinear optical devices by
utilizing the difference in the magnitude of inter-subband
absorption between the TM mode and the TE mode. It may be needless
to say that the first semiconductor optical waveguide layer too is
not limited to InGaAsP type substances. Additionally, a material
other than semiconductor may be inserted between the first and
second semiconductor optical waveguide layers. Materials and
wavelengths other than those specifically described above may be
used in many different combinations.
The buried type, the ridge-mesa type and various other types may be
used for the optical waveguide. The arrangement for confining the
electric current to the active layer and the means for applying an
electric field to the second semiconductor optical waveguide layer
are not limited to those specifically described above by referring
to the embodiments.
The present invention is applicable to a variety of devices other
than those described above by referring to the embodiments and they
include polarized wave control devices (capable of controlling
absorption in the TM mode), wavelength selection type optical
modulators, and optical modulators and amplifiers capable of
controlling the a parameter. A device according to the present
invention and having functional features same as those of its
conventional counterpart may have its functions improved
remarkably. For instance, an optical control type optical switch
according to the present invention can operate highly efficiently
and may be provided with regulation capabilities. Since a
semiconductor optical waveguide device according to the present
invention is a multi-functional device as described earlier by
referring to the eighth embodiment, a highly integrated single chip
device having a variety of functional features may be realized by
using such a semiconductor optical waveguide device.
As described in detail for the eighth and ninth embodiments above,
a semiconductor optical waveguide device according to the present
invention can be used for a variety of new applications including a
rapidly wavelength tunable filter, a high efficiency wavelength
conversion device and a rapidly wavelength tunable laser as well as
an integrated device realized by combining the functional features
of such devices. Thus, the present invention can be used for
realize highly efficient multi-functional light sources, light
receiving devices and optical waveguide devices.
Additional advantages and modifications will readily occur to those
skilled in the art. Therefore, the present invention in its broader
aspects is not limited to the specific details, representative
devices, and illustrated examples shown and described herein.
Accordingly, various modifications may be made without departing
from the spirit or scope of the general inventive concept as
defined by the appended claims and their equivalents.
* * * * *